Argued December 6, 1983 Decided April 18, 1984
Under Massachusetts law, a defendant charged with certain minor crimes in Boston Municipal Court may elect to have a bench trial or a jury trial. If he chooses a jury trial and is convicted, he has the normal appellate process open to him, but if he chooses a bench trial and is dissatisfied with the results, he has an absolute right to a trial de novo before a jury and need not allege error at the bench trial to obtain de novo review. However, there is no right to appellate review of a bench trial conviction. Respondent elected to undergo a first-tier bench trial on a charge of knowing possession of implements designed for breaking into an automobile to steal property. He was convicted and sentenced to a jail term, the trial judge having rejected his claim that the prosecution had introduced no evidence of intent to steal. Respondent then requested a de novo jury trial and was released on personal recognizance pending retrial. Before the jury trial commenced, respondent moved to dismiss the charge on the ground that no evidence of intent had been presented at the bench trial and thus retrial was barred under Burks v. United States, 437 U.S. 1, which held that the Double Jeopardy Clause bars a second trial when a reviewing court reverses a conviction on the ground that the evidence presented at the first trial was legally insufficient. The motion to dismiss was denied, and respondent then sought relief in the Massachusetts Supreme Judicial Court, which ultimately held that Burks was inapplicable because no appellate court had ruled that the evidence was insufficient at respondent's bench trial. The Massachusetts court also ruled that a trial de novo without a determination as to the sufficiency of the evidence at the bench trial would not violate the Double Jeopardy Clause. Respondent then sought habeas corpus relief in Federal District Court, which held that respondent was "in custody" for purposes of 28 U.S.C. § 2254(b) and that he had exhausted his state remedies. Finding for respondent on the merits, the court concluded that, under Burks, a second trial was foreclosed if the evidence against respondent at the bench trial was insufficient, and that there was insufficient evidence of intent at the bench trial to support respondent's conviction. The Court of Appeals affirmed.
1. The District Court had jurisdiction to entertain respondent's habeas corpus action. Pp. 300-303.
(a) For purposes of the federal habeas corpus statutes, respondent was in "custody" even though his conviction was vacated when he applied for a trial de novo and he had been released on personal recognizance. The use of habeas corpus is not restricted to situations in which the applicant is in actual physical custody. The Massachusetts statute under which respondent was released subjected him to restraints not shared by the public generally, including the obligations to appear in court for trial and not to depart without leave. Cf. Hensley v. Municipal Court, 411 U.S. 345. Pp. 300-302.
(b) Respondent had exhausted his state remedies with respect to his double jeopardy claim. The Massachusetts Supreme Judicial Court rejected his claim, and the fact that he might ultimately be acquitted at the trial de novo did not alter the fact that he had taken his claim that he should not be tried again as far as he could in the state courts. A requirement that a defendant run the entire gamut of state procedures, including retrial, prior to consideration of his claim in federal court, would require him to sacrifice the protection of the Double Jeopardy Clause against being twice put to trial for the same offense. Pp. 302-303.
2. Respondent's retrial de novo without any judicial determination of the sufficiency of the evidence at his prior bench trial will not violate the Double Jeopardy Clause. Pp. 304-313.
(a) Ludwig v. Massachusetts, 427 U.S. 618 — upholding a prior Massachusetts two-tier system of trial courts that differed from the present one by requiring a defendant to participate in the first-tier proceedings and by not allowing him to choose a jury trial in the first instance — was not disturbed by the decision in Burks, supra, and is dispositive of the double jeopardy issue here. Pp. 304-306.
(b) In this case, the State is not attempting, contrary to the guarantees embodied in the Double Jeopardy Clause, to impose multiple punishments for a single offense or to convict respondent after acquittal. Respondent has not been acquitted; he simply maintains that he ought to have been. Pp. 306-308.
(c) The concept of "continuing jeopardy" is implicit in the general rule that the Double Jeopardy Clause does not bar retrial after reversal of a conviction. Acquittal terminates the initial jeopardy, and Burks recognizes that a determination by a reviewing court that the evidence was legally insufficient likewise terminates the initial jeopardy. Respondent failed to identify any stage of the state proceedings that can be held to have terminated jeopardy. Pp. 308-310.
(d) The Massachusetts system does not constitute governmental oppression of the sort against which the Double Jeopardy Clause was intended to protect, even when a defendant convicted at the first tier claims insufficiency of the evidence. The defendant's absolute right to obtain a de novo jury trial without alleging error at the bench trial ameliorates the danger of affording the prosecution an opportunity to supply evidence which it failed to muster in the first proceeding. The prosecution has every incentive to put forward its strongest case at the bench trial, because an acquittal would preclude reprosecution of the defendant. There is nothing to stop a defendant from choosing a bench trial for the sole purpose of getting a preview of the State's case to enable him to prepare better for the jury trial. The two-tier system, unlike a more conventional system, gives a defendant two opportunities to be acquitted on the facts. If the prosecution obtains a conviction at the second trial, the defendant then has the usual appellate remedies. Pp. 310-312.
698 F.2d 1, reversed.
WHITE, J., delivered the opinion of the Court, in which BLACKMUN and REHNQUIST, JJ., joined; in Parts I and II of which BRENNAN, MARSHALL, and STEVENS, JJ., joined; and in Parts I, II-B, III, and IV of which BURGER, C.J., and POWELL, J., joined. BRENNAN, J., filed an opinion concurring in part and concurring in the judgment, in which MARSHALL, J., joined, post, p. 313. POWELL, J., filed an opinion concurring in part and concurring in the judgment, in which BURGER, C.J., joined, post, p. 327. STEVENS, J., filed an opinion concurring in part and concurring in the judgment, post, p. 328. O'CONNOR, J., filed an opinion concurring in the judgment, post, p. 337.
Barbara A. H. Smith, Assistant Attorney General of Massachusetts, argued the cause for petitioners. With her on the briefs were Francis X. Bellotti, Attorney General, and Michael J. Traft.
Page 296 Eric D. Blumenson, Burt Neuborne, Charles S. Sims, John Reinstein, and Marjorie Heins filed a brief for the American Civil Liberties Union et al. as amici curiae urging affirmance.
We granted certiorari, 463 U.S. 1206 (1983), to review a decision of the Court of Appeals for the First Circuit affirming the issuance of a writ of habeas corpus. The Court of Appeals agreed with the District Court that the trial de novo of respondent Lydon, pursuant to Massachusetts' "two-tier" system for trying minor crimes, would violate his right not to be placed twice in jeopardy for the same crime, because it determined that insufficient evidence of a critical element of the charge was adduced at the first-tier trial. We reverse.
Under Massachusetts law, a defendant charged with certain crimes in Boston Municipal Court may elect either a bench trial or a jury trial. Mass. Gen. Laws Ann., ch. 218, §§ 26, 26A (West Supp. 1983-1984). If a defendant chooses a jury and is convicted, he has the normal appellate process open to him, while a defendant dissatisfied with the results of a bench trial, if he elects that course, has an absolute right to a trial de novo before a jury. §§ 26 and 27A. A convicted defendant who has chosen a bench trial need not allege error at that trial to obtain de novo review. On the other hand, he may not rely upon error at the bench trial to obtain reversal of his conviction; his only recourse is a trial de novo.
At the second-tier trial, a defendant may waive a jury and undergo a second bench trial. Mass. Gen. Laws Ann., ch. 218, § 27A(g) (West Supp. 1983-1984).
Respondent Michael Lydon was arrested after breaking into an automobile in Boston. He was charged with the knowing possession of implements "adapted and designed for forcing and breaking open a depository [an automobile] in order to steal therefrom, such money or other property as might be found therein" with intent "to use and employ them therefor." Record, Complaint. Lydon elected to undergo a first-tier bench trial and was convicted. The trial judge rejected Lydon's claim that the prosecution had introduced no evidence that Lydon intended to steal from the car and that his actions were as consistent with activities not covered by the complaint. Lydon was sentenced to two years in jail.
Lydon requested a trial de novo in the jury session of the Boston Municipal Court. Pending retrial, he was released on personal recognizance. Before the jury trial commenced, Lydon moved to dismiss the charge against him on the ground that no evidence of the element of intent had been presented at the bench trial. He contended that retrial was therefore barred under the principles of Burks v. United States, 437 U.S. 1 (1978), which held that the Double Jeopardy Clause bars a second trial when a reviewing court reverses a conviction on the ground that the evidence presented at the first trial was legally insufficient.
After the motion to dismiss was denied, Lydon sought relief in the single justice session of the Supreme Judicial Court of Massachusetts. See Mass. Gen. Laws Ann., ch. 211, § 3 (West 1958). The single justice issued a stay of the de novo trial and reported two questions to the full bench:
"1. Is it a denial of a defendant's right not to be placed in double jeopardy to require him to go through a jury trial, requested by him without waiving his rights, when the evidence at the bench trial was insufficient to warrant a conviction?
"2. Assuming that a jury trial in such an instance would be a denial of a defendant's right not to be placed in double jeopardy, may the issue of the sufficiency of the evidence at the bench trial be considered again at the trial court level, assuming, of course, that the judge at the bench trial has denied an appropriate request for a ruling that the evidence at the bench trial was insufficient?"
The single justice did not report a finding on the sufficiency of the evidence, although he did state that he was "of the view that the evidence was not sufficient to warrant guilty findings." Record, Reservation and Report, at 3. He also noted that the prosecution conceded that the evidence presented was insufficient to warrant a finding of guilt on the charges set forth in the complaint. Ibid.
On review by the Supreme Judicial Court, the court initially noted that the single justice did not sit as a reviewing court in determining the sufficiency of the evidence and that any conclusion reached by him on that issue "was made for the purpose of reporting clearly framed questions to the full bench and is not an adjudication of the rights of the parties in this case." Lydon v. Commonwealth, 381 Mass. 356, 359, n. 6, 409 N.E.2d 745, 748, n. 6, cert. denied, 449 U.S. 1065 (1980). The Massachusetts court then found Lydon's double jeopardy argument to be without merit. Because no appellate court had ruled that the evidence was insufficient at Lydon's trial, and indeed no court ever would have occasion to do so under Massachusetts law, the court found Burks inapplicable. Burks, the court observed, did not address the question whether under double jeopardy principles a defendant convicted on insufficient evidence at a bench trial has a right to reconsideration of the sufficiency of the evidence prior to a trial de novo. The court concluded that "[a] defendant is not placed in double jeopardy merely because his only avenue of relief from a conviction based on insufficient evidence at a voluntarily sought bench trial is a trial de novo." 381 Mass., at 367, 409 N.E.2d, at 752. As to the second reported question, the court concluded that if there is a valid double jeopardy claim, it should be dealt with prior to the trial de novo, although it acknowledged that its conclusion on this question was "rendered largely academic" by its answer to the first question since any double jeopardy claim presented to the second-tier court would necessarily be rejected. Id., at 366, 409 N.E.2d, at 752.
Lydon then filed a petition for a writ of habeas corpus in the United States District Court for the District of Massachusetts. First addressing the question of its jurisdiction, the District Court held that Lydon was "in custody" for purposes of 28 U.S.C. § 2254(b) and that he had exhausted his state remedies because there was no state remedy available to him short of submitting to a second trial. 536 F. Supp. 647 (1982). On the merits, the District Court viewed Burks v. United States, supra, as "bestow[ing] a constitutional right upon defendants not to be retried when the initial conviction rests on insufficient evidence," 536 F. Supp., at 651, and thought that this holding foreclosed a second trial if the evidence against Lydon at the bench trial was insufficient, id., at 652. After reviewing the transcript of the bench trial, the District Court concluded that there was insufficient, evidence of intent to support a conviction and ordered the writ to issue. On appeal, a divided Court of Appeals for the First Circuit affirmed in all respects. 698 F.2d 1 (1982).
We first address the Commonwealth's contention that the District Court lacked jurisdiction to entertain Lydon's habeas corpus action because he was not in "custody" for purposes of the statute and had not exhausted his state remedies. Under 28 U.S.C. § 2241(c), a "writ of habeas corpus shall not extend to a prisoner unless . . . (3) He is in custody in violation of the Constitution or laws or treaties of the United States." Similarly, 28 U.S.C. § 2254(a) states that a writ of habeas corpus is available to persons "in custody pursuant to the judgment of a State court." Petitioners argue that because Lydon's first conviction had been vacated when he applied for a trial de novo, and because he had been released on personal recognizance, he was not in "custody."
Our cases make clear that "the use of habeas corpus has not been restricted to situations in which the applicant is in actual, physical custody." Jones v. Cunningham, 371 U.S. 236, 239 (1963). In Hensley v. Municipal Court, 411 U.S. 345 (1973), we held that a petitioner enlarged on his own recognizance pending execution of sentence was in custody within the meaning of 28 U.S.C. § 2241(c)(3) and 2254(a). Hensley's release on personal recognizance was subject to the conditions that he would appear when ordered by the court, that he would waive extradition if he was apprehended outside the State, and that a court could revoke the order of release and require that he be returned to confinement or post bail. Although the restraints on Lydon's freedom are not identical to those imposed on Hensley, we do not think that they are sufficiently different to require a different result.
The Massachusetts statute under which Lydon was released subjects him to "restraints not shared by the public generally." 411 U.S., at 351. He is under an obligation to appear for trial in the jury session on the scheduled day and also "at any subsequent time to which the case may be continued . . . and so from time to time until the final sentence." Mass. Gen. Laws Ann., ch. 278, § 18 (West 1981). Failure to appear "without sufficient excuse" constitutes a criminal offense. Ch. 276, § 82A. Also, if Lydon fails to appear in the jury session, he may be required, without a further trial, to serve the 2-year sentence originally imposed. Ch. 278, § 24. Finally, the statute requires that he "not depart without leave, and in the meantime . . . keep the peace and be of good behavior." Ch. 278, § 18. Consequently, we believe that the Court of Appeals correctly held that Lydon was in custody.
Petitioners contend that a conclusion that a person released on personal recognizance is in custody for purposes of the federal habeas corpus statutes will "ope[n] the door to the federal court to all persons prior to trial." Brief for Petitioners 24. We addressed the same argument in Hensley:
"Finally, we emphasize that our decision does not open the doors of the district courts to the habeas corpus petitions of all persons released on bail or on their own recognizance. We are concerned here with a petitioner who has been convicted in state court and who has apparently exhausted all available state court opportunities to have that conviction set aside. Where a state defendant is released on bail or on his own recognizance pending trial or pending appeal, he must still contend with the requirements of the exhaustion doctrine if he seeks habeas corpus relief in the federal courts. Nothing in today's opinion alters the application of that doctrine to such a defendant." 411 U.S., at 353.
We do not carve out a special-purpose jurisdictional exception for double jeopardy allegations with respect to custody. Nothing in our discussion of custody is dependent upon the nature of the claim that is raised. To the extent that double jeopardy claims are treated differently for habeas purposes, it is because of the application of the exhaustion principle, not because a different definition of custody is adopted.
We are also convinced that Lydon had exhausted his state remedies with respect to his claim that his second trial would violate his right not to be twice placed in jeopardy unless it is judicially determined that the evidence at his first trial was sufficient to sustain his conviction. This precise claim was presented to and rejected by the Supreme Judicial Court of Massachusetts. That court definitively ruled that Lydon had no right to a review of the sufficiency of the evidence at the first trial and that his trial de novo without such a determination would not violate the Double Jeopardy Clause. That Lydon may ultimately be acquitted at the trial de novo does not alter the fact that he has taken his claim that he should not be tried again as far as he can in the state courts.
The exhaustion requirement is set forth in 28 U.S.C. § 2254, which provides in relevant part:
"(b) An application for a writ of habeas corpus in behalf of a person in custody pursuant to the judgment of a State court shall not be granted unless it appears that the applicant has exhausted the remedies available in the courts of the State, or that there is either an absence of available State corrective process or the existence of circumstances rendering such process ineffective to protect the rights of the prisoner.
"(c) An applicant shall not be deemed to have exhausted the remedies available in the courts of the State, within the meaning of this section, if he has the right under the law of the State to raise, by any available procedure, the question presented."
We should keep in mind in this respect the unique nature of the double jeopardy right. In Abney v. United States, 431 U.S. 651 (1977), the Court held that denial of a motion to dismiss an indictment on double jeopardy grounds constitutes a final order for purposes of 28 U.S.C. § 1291. That decision was based upon the special nature of the double jeopardy right and the recognition that the right cannot be fully vindicated on appeal following final judgment, since in part the Double Jeopardy Clause protects "against being twice put to trial for the same offense." Id., at 661 (emphasis in original). Because the Clause "protects interests wholly unrelated to the propriety of any subsequent conviction," ibid., a requirement that a defendant run the entire gamut of state procedures, including retrial, prior to consideration of his claim in federal court, would require him to sacrifice one of the protections of the Double Jeopardy Clause.
Section 2254(b) specifically allows for the issuance of habeas writs when circumstances exist "rendering [state] process ineffective to protect the rights of the prisoner." In the circumstances of this case, there are no more state procedures of which Lydon may avail himself to avoid an allegedly unconstitutional second trial.
In our view, therefore, Lydon had exhausted his double jeopardy claim in the state courts, and that precondition to the District Court's jurisdiction was satisfied. We conclude below, however, that the District Court and the Court of Appeals erred in sustaining Lydon's double jeopardy claim: in our view, Lydon could be retried de novo without any judicial determination of the sufficiency of the evidence at his prior bench trial.
If our conclusion were otherwise, a further exhaustion issue would arise. The District Court and the Court of Appeals not only held that Lydon was entitled to a determination of the sufficiency of the evidence at his first trial but also proceeded to make this evidentiary determination. Yet it seems to us that the Supreme Judicial Court of Massachusetts held that any double jeopardy claim Lydon might have should be made prior to the beginning of the second trial, although it candidly stated that under its opinion no such claim could succeed. If the Massachusetts court was wrong, however, in ruling that Lydon was not entitled to a sufficiency determination, it is apparent that the way would be open for him to present his claim to the de novo court in precisely the manner that the Massachusetts court suggested that a double jeopardy claim should be submitted. In our view, therefore, the federal habeas corpus court in any event should not itself have ruled on the sufficiency of the evidence at Lydon's first trial Page 304 but should have stayed its hand and permitted the state court to make that determination in the first instance. Otherwise, Lydon could not be said to have exhausted his state remedies and satisfied the requirements of § 2254.
It is for that reason that reliance by Lydon and the courts below on Jackson v. Virginia, 443 U.S. 307 (1979), is misplaced. Jackson held that federal habeas courts must consider a petitioner's federal due process claim that the evidence in support of his conviction was insufficient to have led a rational trier of fact to find him guilty beyond a reasonable doubt. No one has suggested, however, that Jackson in any way created an exception to the exhaustion requirement.
Because in our view Lydon may be retried and convicted without a review of the sufficiency of the evidence at his bench trial, there will never be an occasion for a federal habeas corpus court to deal with the evidentiary issue at that trial. Since JUSTICE STEVENS disagrees with our double jeopardy decision, he asserts that the federal court must perform its Jackson v. Virginia function with respect to the evidence at the first trial. He would postpone that task until after the second trial, however. Of course, if Lydon is convicted at his jury trial, the sufficiency of the evidence at that trial will concededly be open to review in a federal court, as Jackson v. Virginia mandates.
In Ludwig v. Massachusetts, 427 U.S. 618 (1976), we upheld a prior Massachusetts two-tier system of trial courts for criminal cases. The present system differs from the system upheld in Ludwig in only one respect of significance here. Prior to the Massachusetts Court Reorganization Act of 1978, a defendant could not elect a jury trial in the first instance; he was required to participate in the first-tier proceedings. Under the present system, as noted above, a defendant may avoid the first-tier trial altogether and proceed directly to the jury trial. In upholding the prior Massachusetts system, we stated:
"The Massachusetts system presents no danger of prosecution after an accused has been pardoned; nor is there any doubt that acquittal at the first tier precludes reprosecution. Instead, the argument appears to be that because the appellant has been placed once in jeopardy and convicted, the State may not retry him when he informs the trial court of his decision to `appeal' and to secure a trial de novo.
"Appellant's argument is without substance. The decision to secure a new trial rests with the accused alone. A defendant who elects to be tried de novo in Massachusetts is in no different position than is a convicted defendant who successfully appeals on the basis of the trial record and gains a reversal of his conviction and a remand of his case for a new trial. Under these circumstances, it long has been clear that the State may reprosecute. United States v. Ball, 163 U.S. 662 (1896). The only difference between an appeal on the record and an appeal resulting automatically in a new trial is that a convicted defendant in Massachusetts may obtain a `reversal' and a new trial without assignment of error in the proceedings at his first trial. Nothing in the Double Jeopardy Clause prohibits a State from affording a defendant two opportunities to avoid conviction and secure an acquittal." Id., at 631-632.
Our decision in Ludwig, which we think is dispositive of the double jeopardy issue in this case, was not disturbed by our later decision in Burks v. United States, 437 U.S. 1 (1978). In Burks, the petitioner's conviction had been set aside by the Court of Appeals on the ground that there had been insufficient evidence presented at his trial to support the verdict. The Court of Appeals then ordered the case remanded to the District Court for a determination of whether a new trial should be ordered or a directed verdict of acquittal should be entered. We reversed, stating:
"In short, reversal for trial error, as distinguished from evidentiary insufficiency, does not constitute a decision to the effect that the government has failed to prove its case. As such, it implies nothing with respect to the guilt or innocence of the defendant. . . .
"The same cannot be said when a defendant's conviction has been overturned due to a failure of proof at trial, in which case the prosecution cannot complain of prejudice, for it has been given one fair opportunity to offer whatever proof it could assemble. Moreover, such an appellate reversal means that the government's case was so lacking that it should not have even been submitted to the jury. Since we necessarily afford absolute finality to a jury's verdict of acquittal — no matter how erroneous its decision — it is difficult to conceive how society has any greater interest in retrying a defendant when, on review, it is decided as a matter of law that the jury could not properly have returned a verdict of guilty." Id., at 15-16. (footnote omitted) (emphasis in original).
We summarized our holding in Burks as being "that the Double Jeopardy Clause precludes a second trial once the reviewing court has found the evidence legally insufficient." Id., at 18.
Lydon argues, and the Court of Appeals held, that our statement in Ludwig that a defendant who elects to be tried de novo is in the same position as a convicted defendant who successfully appeals, combined with our holding in Burks that the setting aside of a conviction on the basis of evidentiary insufficiency bars retrial, mandates the conclusion that a trial de novo is barred by the Double Jeopardy Clause if the evidence presented at the bench trial was insufficient to support a finding of guilt. We are unpersuaded.
The Double Jeopardy Clause of the Fifth Amendment provides that no person shall "be subject for the same offence to be twice put in jeopardy of life or limb." In Benton v. Maryland, 395 U.S. 784 (1969), we held that this guarantee is applicable to the States through the Fourteenth Amendment.
Our cases have recognized three separate guarantees embodied in the Double Jeopardy Clause: It protects against a second prosecution for the same offense after acquittal, against a second prosecution for the same offense after conviction, and against multiple punishments for the same offense. Illinois v. Vitale, 447 U.S. 410, 415 (1980). The primary goal of barring reprosecution after acquittal is to prevent the State from mounting successive prosecutions and thereby wearing down the defendant. As was explained in Green v. United States, 355 U.S. 184, 187-188 (1957):
The Clause also, of course, protects against retrial after the declaration of a mistrial in certain circumstances. See United States v. Scott, 437 U.S. 82 (1978).
"The underlying idea, one that is deeply ingrained in at least the Anglo-American system of jurisprudence, is that the State with all its resources and power should not be allowed to make repeated attempts to convict an individual for an alleged offense, thereby subjecting him to embarrassment, expense and ordeal and compelling him to live in a continuing state of anxiety and insecurity, as well as enhancing the possibility that even though innocent he may be found guilty."
The primary purpose of foreclosing a second prosecution after conviction, on the other hand, is to prevent a defendant from being subjected to multiple punishments for the same offense. See United States v. Wilson, 420 U.S. 332, 343 (1975).
In this case, the Commonwealth is not attempting to impose multiple punishments for a single offense. Nor is it making another attempt to convict Lydon after acquittal. It is satisfied with the results of the bench trial and would have abided the results of a jury trial had Lydon taken that initial course. The conceptual difficulty for Lydon is that he has not been acquitted; he simply maintains that he ought to have been. His claim is that the evidence at the bench trial was insufficient to convict and that a second trial to a jury will offend the fundamental rule that a verdict of acquittal may "not be reviewed, on error or otherwise, without putting [a defendant] twice in jeopardy." United States v. Ball, 163 U.S. 662, 671 (1896); United States v. Martin Linen Supply Co., 430 U.S. 564, 571 (1977). Our cases, however, do not take us as far as Lydon would like.
The Double Jeopardy Clause is not an absolute bar to successive trials. The general rule is that the Clause does not bar reprosecution of a defendant whose conviction is overturned on appeal. United States v. Ball, supra. The justification for this rule was explained in United States v. Tateo, 377 U.S. 463, 466 (1964), as follows:
"While different theories have been advanced to support the permissibility of retrial, of greater importance than the conceptual abstractions employed to explain the Ball principle are the implications of that principle for the sound administration of justice. Corresponding to the right of an accused to be given a fair trial is the societal interest in punishing one whose guilt is clear after he has obtained such a trial. It would be a high price indeed for society to pay were every accused granted immunity from punishment because of any defect sufficient to constitute reversible error in the proceedings leading to conviction."
In Price v. Georgia, 398 U.S. 323, 329 (1970), we recognized that implicit in the Ball rule permitting retrial after reversal of a conviction is the concept of "continuing jeopardy." See also Breed v. Jones, 421 U.S. 519, 534 (1975). That principle "has application where criminal proceedings against an accused have not run their full course." 398 U.S., at 326. Interests supporting the continuing jeopardy principle involve fairness to society, lack of finality, and limited waiver. Id., at 329, n. 4. Acquittals, unlike convictions, terminate the initial jeopardy. This is so whether they are "express or implied by a conviction on a lesser included offense." Id., at 329. In Burks, 437 U.S. 1 (1978), we recognized that an unreversed determination by a reviewing court that the evidence was legally insufficient likewise served to terminate the initial jeopardy.
We assume, without deciding, that jeopardy attached at the swearing of the first witness at Lydon's bench trial. The question then is whether jeopardy has now terminated. Lydon's double jeopardy argument requires an affirmative answer to that question, but he fails to identify any stage of the state proceedings that can be held to have terminated jeopardy. Unlike Burks, who could rest his claim upon the appellate court's determination of insufficiency, Lydon is faced with the unreversed determination of the bench-trial judge, contrary to Lydon's assertion, that the prosecution had met its burden of proof. We noted in United States v. Martin Linen Supply Co., supra, at 571, that an acquittal "represents a resolution, correct or not, of some or all of the factual elements of the offense charged." (Emphasis added.) Lydon's claim of evidentiary failure and a legal judgment to that effect therefore have different consequences under the Double Jeopardy Clause. We believe that the dissent in the Court of Appeals correctly described the nature of the de novo hearing as follows:
"While technically [the defendant] is `tried again,' the second stage proceeding can be regarded as but an enlarged, fact-sensitive part of a single, continuous course of judicial proceedings during which, sooner or later, a defendant receives more — rather than less — of the process normally extended to criminal defendants in this nation." 698 F.2d, at 12 (Campbell, J., dissenting).
In Burks, the question involved the significance to be attached to a particular event — an appellate determination that the evidence was insufficient to support a conviction. Concededly, no such event has occurred here; but Lydon insists that he is entitled under the Federal Constitution to a review of the evidence presented at the bench trial before proceeding with the second-tier trial. Burks does not control this very different issue, and we are convinced that the Double Jeopardy Clause does not reach so far. Consequently, we reject the suggestion that Burks modified Ludwig, and we reaffirm our holding in the latter case.
JUSTICE BRENNAN suggests that the voluntary nature of the two-tier system strongly influences his conclusion. Post, at 325-326, and n. 8. It is not clear why that is so, given that his reasoning is based upon the defendant's expectations, rather than a theory of waiver.
A number of features of the Massachusetts system persuade us that it does not constitute "governmental oppression of the sort against which the Double Jeopardy Clause was intended to protect," United States v. Scott, 437 U.S. 82, 91 (1978), even when a defendant convicted at the first tier claims insufficiency of the evidence.
We note at the outset that Lydon was in "jeopardy" in only a theoretical sense. Although technically "jeopardy" under the Double Jeopardy Clause entails the "potential or risk of trial and conviction, not punishment," Price v. Georgia, supra, at 329, it is worthy of note that virtually nothing can happen to a defendant at a first-tier trial that he cannot avoid. He has an absolute right to obtain the de novo trial, and he need not allege error at the first-tier trial to do so. Once the right to a de novo trial is exercised, the judgment at the bench trial is "wiped out." Mann v. Commonwealth, 359 Mass. 661, 271 N.E.2d 331 (1971).
The defendant's right to obtain de novo review without alleging error is significant in that it ameliorates one of the concerns underlying our opinion in Burks. In Burks, we recognized the danger of "affording the prosecution another opportunity to supply evidence which it failed to muster in the first proceeding." 437 U.S., at 11. The Court of Appeals in this case stated that "[t]he process of judicial review has conveniently pinpointed the evidence which was lacking, and retrial simply gives the prosecutor another opportunity to supply it." 698 F.2d, at 8. However, the "process of judicial review" that resulted in the identification of the precise area of insufficiency is not a part of the ordinary Massachusetts procedure and would not have occurred had it not been for Lydon's double jeopardy claim and the intervention by federal courts. In the usual case, there would be no review prior to the jury trial.
A claim that our decision in this case creates an incentive for a prosecutor to hold back and learn the defendant's case in the first trial, in order to hone his presentation in the second, is unpersuasive. The prosecution has every incentive to put forward its strongest case at the bench trial, because an acquittal will preclude reprosecution of the defendant. Although admittedly the Commonwealth at the de novo trial will have the benefit of having seen the defense, the defendant likewise will have had the opportunity to assess the prosecution's case. Because in most cases the judge presiding at the bench trial can be expected to acquit a defendant when legally insufficient evidence has been presented, it is clear that the system provides substantial benefits to defendants, as well as to the Commonwealth. In fact, as we recognized in Ludwig v. Massachusetts, 427 U.S., at 626-627, there appears to be nothing to stop a defendant from choosing a bench trial for the sole purpose of getting a preview of the Commonwealth's case to enable him to prepare better for the jury trial. To put the matter another way, as we observed in Colten v. Kentucky, 407 U.S. 104, 119 (1972), a defendant's chances in a two-tier system are "[i]n reality . . . to accept the decision of the judge and the sentence imposed in the inferior court or to reject what in effect is no more than an offer in settlement of his case and seek the judgment of a judge or jury in the superior court, with sentence to be determined by the full record made in that court."
It appears that defendants recognize the advantages of two-tier systems. During one period studied, only about 9% of defendants chose a jury trial in the first instance. Moreover, thousands of cases were disposed of by convictions at bench trials because many convicted defendants did not exercise their right to appeal to the jury trial session. Lydon v. Commonwealth, 381 Mass. 356, 359, n. 5, 409 N.E.2d 745, 748, n. 5, cert. denied, 449 U.S. 1065 (1980).
We also note the fact that the advantages of two-tier systems have led almost half of the States to adopt such systems. See 698 F.2d 1, 2 (CA1 1982).
As the dissent in the Court of Appeals recognized, the two-tier system affords benefits to defendants that are unavailable in a more conventional system. 698 F.2d, at 11-12 (Campbell, J., dissenting). In traditional systems, a convicted defendant may seek reversal only on matters of law; in the Massachusetts system a defendant is given two opportunities to be acquitted on the facts. If he is acquitted at the first trial, he cannot be retried. See Ludwig v. Massachusetts, supra, at 631. If he is convicted, he may then choose to invoke his right to a trial de novo and once again put the prosecution to its proof. If the prosecution fails in the second trial to convince the trier-of-fact of the defendant's guilt beyond a reasonable doubt, an acquittal results. If the prosecution succeeds in obtaining a conviction the second time, the defendant then has the usual appellate remedies. As we noted in Ludwig, "[n]othing in the Double Jeopardy Clause prohibits a State from affording a defendant two opportunities to avoid conviction and secure an acquittal." 427 U.S., at 632.
Of course, under the present Massachusetts two-tier system, a defendant can also wholly avoid the consequences of a first-tier trial by avoiding the trial altogether. A defendant has an unqualified right to proceed to a jury trial in the first instance. It thus cannot be said that the Commonwealth required that Lydon submit to two trials. In this sense, the current Massachusetts system is more favorable to defendants than was the system we upheld against constitutional attack in Ludwig v. Massachusetts. There is not the slightest hint in the record that Lydon, who was represented by counsel, did not choose the bench trial voluntarily.
Although, as Judge Campbell said in dissent below, his colleagues' opinion reflects "intelligence and logic," we agree with him that their "relentless application of secondary precepts developed in other, very different settings" led to a wrong result not required by the Constitution and destructive of "a useful and fair state procedure." 698 F.2d, at 10. Accordingly, we reverse the judgment of the Court of Appeals.
I agree that, because respondent was "in custody" within the meaning of 28 U.S.C. § 2241(c)(3) and 2254(a) and because he had exhausted all available state remedies for his constitutional claim, the District Court had jurisdiction to entertain his habeas corpus petition. Accordingly, I join Parts I and II of the Court's opinion. I analyze the merits differently than does the Court, however, and therefore do not join Parts III and IV of its opinion.
Although it appears in Part II in which I otherwise concur, I do not agree with the implications of footnote 5 of the Court's opinion. See n. 7, infra.
The Court rejects Lydon's double jeopardy claim by relying on the absence of "government oppression" and the presence of "continuing jeopardy." For many of the reasons advanced by the Court, as well as others, see infra, at 324-326, I completely agree that the two-tier trial option available to Massachusetts defendants appears eminently fair and reasonable and that there is therefore no evidence of the kind of "governmental oppression" that might, apart from other analytical considerations, provide an independent basis for a double jeopardy claim. I do not, however, believe — nor do I understand the Court to suggest — that the absence of "governmental oppression" standing alone would defeat a double jeopardy claim otherwise valid under our cases.
At first blush, Lydon appears to present such a claim. The Court assumes, as petitioners concede, "that jeopardy attached at the swearing of the first witness at Lydon's bench trial," ante, at 309; the Commonwealth does not claim it lacked "a fair opportunity" to present its best evidence nor does it challenge the District Court's determination, based on an application of Massachusetts decisions directly on point, that "the State had failed as a matter of law to prove its case" against Lydon, see 698 F.2d 1, 7 (CA1 1982) (opinion below); and, finally, the Court seems to acknowledge that, as a result of today's decision, Lydon will undergo two trials, ante, at 309. Accordingly, Lydon appears to establish that, contrary to the rule we unanimously reaffirmed just three Terms ago, he will be subjected to "retrial where the State has failed as a matter of law to prove its case despite a fair opportunity to do so." Hudson v. Louisiana, 450 U.S. 40, 45, n. 5 (1981).
Lydon does not contend that the Commonwealth is required by the Federal Constitution to afford appellate review of the evidence presented at the bench trial before proceeding with the second-tier trial. See Brief for Respondent 85-90. Instead, Lydon argues that the Commonwealth violated Burks v. United States, 457 U.S. 1 (1978), by ordering him to undergo a second trial, despite what he claims was insufficient evidence at the first trial. As the Court appears to recognize, the jurisdiction of a federal habeas court to entertain such a claim does not depend on the Commonwealth's failure to provide appellate or indeed any other kind of review of the sufficiency before the second trial. The habeas court has jurisdiction as long as the defendant has exhausted whatever state remedies are in fact available. See ante, at 302-303.
The Court meets this argument by noting that Lydon has only a "claim of evidentiary failure . . . [, not] a legal judgment to that effect . . . ." Ante, at 309. Invoking the concept of "continuing jeopardy," the Court maintains that such a "legal judgment" is required before jeopardy is "terminated" and a retrial barred. Nor, in the Court's view, is it enough for these purposes that Lydon has obtained a "legal judgment" that the evidence was constitutionally inadequate from a Federal District Court, acting within its jurisdiction and after the defendant has exhausted state remedies. Instead, Lydon's claim must be rejected because "he fails to identify any stage of the state proceedings that can be held to have terminated jeopardy." Ante, at 309.
I agree that a valid double jeopardy claim presupposes some identifiable point at which a first trial may be said to have ended. See infra, at 320. I respectfully suggest, however, that mere incantation of the phrase "continuing jeopardy," without more, partakes of the sort of "conceptual abstractions" that our decisions elaborating the requirements of the Double Jeopardy Clause have attempted to avoid. See United States v. Tateo, 377 U.S. 463, 466 (1964). For example, although the Court holds that the Double Jeopardy Clause bars retrial after certain jeopardy-terminating "legal judgments," its approach sets no apparent limits on a State's ability to withhold the necessary "legal judgment," thereby maintaining a state of "continuing jeopardy" and justifying repeated attempts to gain a conviction. And by ignoring the realities of Lydon's situation and demanding a state-court "legal judgment" of acquittal, the Court manages to avoid grappling with the common-sense intuition that the guilty verdict rendered at the end of Lydon's first-tier trial constitutes an obvious point at which proceedings against him "terminated."
Ultimately, the Court's decision rests on an ipse dixit that "[a]cquittals, unlike convictions, terminate the initial jeopardy." Ante, at 308. The Court nowhere explains why an acquittal marks the end of a trial while a conviction or, as in this case, a judgment that the defendant was entitled to an acquittal, lack that effect. Cf. Green v. United States, 355 U.S. 184, 187 (1957), quoting Ex parte Lange, 18 Wall. 163, 169 (1974) ("The common law not only prohibited a second punishment for the same offence, but it went further and forb[ade] a second trial for the same offence, whether the accused had suffered punishment or not, and whether in the former trial he had been acquitted or convicted"). Cf. post, at 329-330 (STEVENS, J., concurring Page 316 in part and concurring in judgment); infra, at 323-327. In any event, if in fact convictions do not terminate jeopardy, then renewed prosecution of a defendant after an unreversed conviction for the same offense — which the Court acknowledges is barred, ante, at 306-307 — would constitute only "continuing" and not double jeopardy under the Court's theory. Nor, under the Court's approach, could the prohibition against such a prosecution be justified by the policy against subjecting a defendant to multiple punishments for the same offense. If a guilty verdict does not "terminate" proceedings, a convicted defendant subjected to further prosecution for the same offense is simply not " twice put in jeopardy" within the language of the Double Jeopardy Clause. U.S. Const., Amdt. 5 (emphasis added). See Missouri v. Hunter, 459 U.S. 359, 366 (1983) ("With respect to cumulative sentences imposed in a single trial, the Double Jeopardy Clause does no more than prevent the sentencing court from prescribing greater punishment than the legislature intended").
To the best of my knowledge, this is the first occasion on which the Court has employed the "continuing jeopardy" notion in such a formalistic fashion. Until today, we have repeatedly emphasized that the concept of "continuing jeopardy" is, at best, a label that "has occasionally been used to explain why an accused who has secured the reversal of a conviction on appeal may be retried for the same offense." Breed v. Jones, 421 U.S. 519, 534 (1975). See also Burks v. United States, 437 U.S. 1, 15 (1978); Price v. Georgia, 398 U.S. 323, 329, n. 4 (1970). But as a talismanic substitute for analysis, the "continuing jeopardy" concept "has `never been adopted by a majority of this Court,'" Breed v. Jones, supra, at 534, quoting United States v. Jenkins, 420 U.S. 358, 369 (1975).
In particular, the rule allowing retrials after reversal for trial error, first announced in United States v. Ball, 163 U.S. 662, 672 (1896), has never rested on the theory that, notwithstanding a guilty verdict ending trial level proceedings, the trial never "terminated" and the defendant therefore remained in a state of "continuing jeopardy." Instead, we have grounded the Ball rule in "the implications of that principle for the sound administration of justice." United States v. Tateo, supra, at 466. See also Tibbs v. Florida, 457 U.S. 31, 40 (1982); United States v. DiFrancesco, 449 U.S. 117, 131 (1980); United States v. Scott, 437 U.S. 82, 89-92 (1978); United States v. Wilson, 420 U.S. 332, 343-344, n. 11 (1975). The opinion in Burks provided the fullest explanation for the Ball rule and also explained why that rule does not permit retrials after reversals based on insufficient evidence:
The Court finds authority for its approach in the statement in Price v. Georgia, 398 U.S. 323, 329 (1970), that "[t]he concept of continuing jeopardy [is] implicit in the Ball case." The opinion in Price did not, however, approve the "broad continuing jeopardy approach," id., at 328, n. 3. Indeed, as the Court notes, ante, at 308, Price suggested that, in light of modern double jeopardy cases, the conclusion represented by the "continuing jeopardy" label reflects "an amalgam of interests — e. g., fairness to society, lack of finality, and limited waiver, among others." 398 U.S., at 329, n. 4. Like Tateo, Jenkins, Breed, Burks, Scott, Wilson, DiFrancesco, and Tibbs, therefore, Price eschewed reliance on the mere shibboleth of "continuing jeopardy."
"[R]eversal for trial error, as distinguished from evidentiary insufficiency, does not constitute a decision to the effect that the government has failed to prove its case. As such, it implies nothing with respect to the guilt or innocence of the defendant. Rather, it is a determination that a defendant has been convicted through a judicial process which is defective in some fundamental respect . . . . When this occurs, the accused has a strong interest in obtaining a fair readjudication of his guilt free from error, just as society maintains a valid concern for insuring that the guilty are punished. . . .
"The same cannot be said when a defendant's conviction has been overturned due to a failure of proof at trial, in which case the prosecution cannot complain of prejudice, for it has been given one fair opportunity to offer whatever proof it could assemble. Moreover, such an appellate reversal means that the government's case was so lacking that it should not have even been submitted to the jury. Since we necessarily afford absolute finality to a jury's verdict of acquittal — no matter how erroneous its decision — it is difficult to conceive how society has any greater interest in retrying a defendant when, on review, it is decided as a matter of law that the jury could not properly have returned a verdict of guilty." Burks v. United States, supra, at 15-16 (emphasis in original) (footnote omitted).
The decision in Burks, therefore, is not merely an application of an abstract concept of "continuing jeopardy." Instead, Burks derives from "[p]erhaps the most fundamental rule in the history of double jeopardy jurisprudence" — that a "`verdict of acquittal . . . [can]not be reviewed, on error or otherwise, without putting [a defendant] twice in jeopardy, and thereby violating the Constitution.'" United States v. Martin Linen Supply Co., 430 U.S. 564, 571 (1977) (quoting United States v. Ball, supra, at 671). Unlike a reversal for trial error, a reversal for constitutionally insufficient evidence represents a determination that, notwithstanding the verdict to the contrary, no "rational trier of fact could have found the essential elements of the crime beyond a reasonable doubt," Jackson v. Virginia, 443 U.S. 307, 319 (1979), and therefore the defendant was entitled to a judgment of acquittal as a matter of law. In the eyes of the law, the defendant is innocent of the charges brought against him. The policies barring retrial after acquittal are no less applicable to such a defendant simply because he, unlike a defendant who actually obtained a judgment of acquittal, was tried before an irrational or lawless factfinder.
To be sure, the Burks rule is not engaged unless the conviction at the first trial is reversed and the State seeks a retrial; Burks forbids a retrial under those circumstances if the evidence at the first trial was constitutionally insufficient. In that respect, the Court is quite correct in stating that a prerequisite to a successful Burks claim is a "legal judgment" rendered at some point that the evidence was insufficient under the standards of Jackson v. Virginia, supra. But the Court's "continuing jeopardy" concept begs the questions of whether and when the defendant is entitled to a judgment barring further proceedings. For all that concept provides, the defendant in Burks was simply fortunate that the reviewing court chose to provide him with a judicial "determination that the evidence was insufficient to support a conviction," ante, at 309, and did not instead rely on an alternative ground of reversal. In the latter event, Burks, like Lydon, would have been left with only a "claim of evidentiary failure[, not] a legal judgment to that effect." Ibid. I cannot agree that the protections of the Double Jeopardy Clause depend so heavily on the grace of a reviewing court. See infra, at 320-321.
Our modern double jeopardy cases have emphasized that, absent substantial countervailing state interests such as ordinarily obtain when a conviction is reversed on grounds of trial error, "the State with all its resources and power should not be allowed to make repeated attempts to convict an individual for an alleged offense, thereby subjecting him to embarrassment, expense and ordeal and compelling him to live in a continuing state of anxiety and insecurity, as well as enhancing the possibility that even though innocent he may be found guilty." Green v. United States, 355 U.S. 184, 187-188 (1957). See also Tibbs v. Florida, 457 U.S. 31, 39-42 (1982). Although the Court quotes the same language from Green, ante, at 307, the "continuing jeopardy" concept on which it relies, as originally set out by Justice Holmes in his dissenting opinion in Kepner v. United States, 195 U.S. 100, 134-137 (1904), entails no discernible limit on the government's ability repeatedly to retry a defendant for "the same cause":
"[I]t seems to me that logically and rationally a man cannot be said to be more than once in jeopardy in the same cause, however often he may be tried. The jeopardy is one continuing jeopardy from its beginning to the end of the cause."
For these reasons, I do not find invocation of an unadorned "continuing jeopardy" concept helpful in resolving the issues posed by this case. Instead, if we are to employ the label "continuing jeopardy," I would attempt to give it content by turning to the principles and policies of the Double Jeopardy Clause that this Court has elaborated in analogous cases.
In order "to be twice put in jeopardy of life or limb" for the same offense, U.S. Const., Amdt. 5 (emphasis added), a defendant facing a new trial must have been subjected to a previous proceeding at which jeopardy attached as a matter of federal constitutional law, Crist v. Bretz, 437 U.S. 28 (1978), and which has now somehow ended; in the Court's terminology, former jeopardy must have "terminated." Of course, it is not sufficient that the defendant claims that one proceeding has concluded and another has begun. For example, the second half of a trial does not subject a defendant to double jeopardy because his motion for a mistrial was denied in the middle of proceedings — even though the defendant asserts that, as far as he is concerned, his trial has ended. Instead, every valid double jeopardy claim presupposes some kind of predicate set of circumstances — such as those typically attendant to a verdict, judgment, or order dismissing the case — objectively concluding one trial and giving rise to the prosecution's effort to begin another.
The question of whether jeopardy has objectively "terminated" should be analyzed in terms of the policies underlying the Double Jeopardy Clause, namely, its concern that repeated trials may subject a defendant "to embarrassment, expense and ordeal and compe[l] him to live in a continuing state of anxiety and insecurity, as well as enhancing the possibility that even though innocent he may be found guilty." Green v. United States, 355 U.S. 184, 187-188 (1957). Jeopardy may be said to have terminated only when the posture of a trial in some objective sense leaves the defendant in such a position that resumption of proceedings would implicate those policies.
Hence, although in most instances a "legal judgment" undoubtedly entails the kind of circumstances under which we may easily conclude that jeopardy has terminated, it seems obvious that a State may not evade the strictures of the Clause simply by withholding a legal judgment and thereby subjecting a defendant to retrial on the theory of "continuing jeopardy." To take two extreme examples, a trial judge, having received a jury verdict of not guilty, may not justify an order that the trial be repeated by refusing to enter a formal judgment on the jury's verdict; nor may a State with a one-tier system avoid a double jeopardy claim by refusing to acknowledge that the first trial had in fact begun and ended. These hypothetical situations, while admittedly unrealistic, nevertheless demonstrate that the determination of whether a trial has in fact "terminated" for purposes of the Double Jeopardy Clause — like the question of whether a trial has begun, Crist v. Bretz, supra — is an issue of federal constitutional law; it cannot turn solely on whether the State has entered a "legal judgment" ending the proceedings. Cf. United States v. Martin Linen Supply Co., 430 U.S., at 571 ("what constitutes an `acquittal' is not to be controlled by the form of the judge's action").
The fact that a trial has ended does not, however, complete the constitutional inquiry; the Court has concluded, most notably in applying the Ball rule, that strong policy reasons may justify subjecting a defendant to two trials in certain circumstances notwithstanding the literal language of the Double Jeopardy Clause. See n. 5, supra. The issue of whether policy reasons of that kind justify retrial in a given case is, however, analytically distinct from the question of whether the challenged proceeding constitutes a second trial or, instead, a continuation of the first. Cases applying the Ball rule, for instance, acknowledge that the defendant will be subjected to two trials but find that fact constitutionally permissible. E. g., United States v. Tateo, 377 U.S., at 465-466.
Accordingly, once it has been determined that a trial has ended as a matter of constitutional law, a court considering a double jeopardy claim must consider the separate question of whether a second trial would violate the Constitution. For example, when a defendant challenging his conviction on appeal contends both that the trial was infected by error and that the evidence was constitutionally insufficient, the court may not, consistent with the rule of Burks v. United States, 437 U.S. 1 (1978), ignore the sufficiency claim, reverse on grounds of trial error, and remand for retrial. Because the first trial has plainly ended, "retrial is foreclosed by the Double Jeopardy Clause if the evidence fails to satisfy the [constitutional standard for sufficiency]. Hence, the [sufficiency] issue cannot be avoided; if retrial is to be had, the evidence must be found to be legally sufficient, as a matter of federal law, to sustain the jury verdict." Tibbs v. Florida, 457 U.S., at 51 (WHITE, J., dissenting). See id., at 45 (majority opinion) (noting that consideration of evidentiary sufficiency before ordering retrial is part of state appellate court's "obligations to enforce applicable state and federal laws").
In short, I believe there are two distinct limitations on a State's ability to retry a defendant on a claim of "continuing jeopardy." First, the issue of whether a trial has ended so that a second trial would constitute double jeopardy is a federal constitutional question, informed but not controlled by the State's characterization of the status of the proceedings; resolution of that question turns essentially on the relationship between the circumstances at issue and the policies underlying the Double Jeopardy Clause. Second, once it has been determined that a first trial has in fact ended, terminating former jeopardy as a matter of federal constitutional law, a State may not place the defendant in jeopardy a second time if retrial is constitutionally barred on any grounds properly preserved and presented.
The approach I have proposed is fully consistent with Ludwig v. Massachusetts, 427 U.S. 618 (1976), and indeed avoids the tension suggested in the Court's opinion between that case and Burks v. United States, 437 U.S. 1 (1978). See ante, at 305-306 and 309-310. As the Court notes, the opinion in Ludwig analogized the second-tier of trial proceedings in Massachusetts to a retrial after reversal of the conviction permissible under the Ball rule. 427 U.S., at 631-632. The Court did not rely on the Page 323 notion that jeopardy continued through both proceedings, rendering them a single "trial," but rather assumed, as in Ball itself, that the second tier constituted a "new trial." 427 U.S., at 632. There was, of course, no suggestion in Ludwig that such a "new trial" was barred because of the absence of constitutionally sufficient evidence — the issue presented by this case — and therefore the Court had no occasion to consider whether the guilty verdict at Ludwig's first-tier trial "terminated" jeopardy.
In this case, the guilty verdict rendered by the first-tier judge undeniably ended a set of proceedings in that courtroom that would be most naturally understood as a single, completed trial. Arguably, therefore, that verdict "terminated" jeopardy. If so, and if the evidence at the first trial was insufficient, then retrial of Lydon at the second tier would be constitutionally barred under Burks, without regard to whether the vacating of the guilty verdict, in and of itself, would otherwise permit a new trial under the Ball rule. And because Lydon has fully exhausted available state remedies, the federal habeas court would be fully authorized to vindicate his claim before trial or after conviction. See ante, at 302-303; Arizona v. Washington, 434 U.S. 497 (1978).
Contrary to the Court's suggestion, Lydon has exhausted every available state remedy for each element of his Burks argument, including that argument's predicate claim that the evidence at the first trial was insufficient. In implying that the sufficiency issue is unexhausted because Lydon failed "to present his claim to the de novo court in precisely the manner that the Massachusetts court suggested that a double jeopardy claim should be submitted," ante, at 303, n. 5, the Court ignores its own earlier statement that "[b]efore the jury trial commenced, Lydon moved to dismiss the charge against him on the ground that no evidence of the element of intent had been presented at the bench trial," ante, at 298. Indeed, the very opinion of the Massachusetts Supreme Judicial Court announcing the proper procedure noted that Lydon had moved to dismiss the case on double jeopardy grounds before the de novo court, Lydon v. Commonwealth, 381 Mass. 356, 357, 366-367, 409 N.E.2d 745, 747, 752 (1980), and, on a petition for review of the jury-trial judge's denial of that motion, agreed that "the jury trial session is the appropriate forum for consideration of double jeopardy claims asserted after a bench trial." Id., at 366-367, 409 N.E.2d, at 752. Accordingly, the Court's effort to avoid the conclusion that Jackson v. Virginia, 443 U.S. 307 (1979), authorized the federal habeas court to consider the sufficiency of the evidence at Lydon's first trial is unavailing.
In the unique context of the Massachusetts two-tier trial system, however, I do not believe a guilty verdict at the first tier is attended by the type of circumstances that can be said to "terminate" trial-level proceedings against Lydon for purposes of the Double Jeopardy Clause. In terms of the policies advanced by the Clause, that verdict has substantially less significance for the defendant than it would have in a traditional, one-tier system. See generally Colten v. Kentucky, 407 U.S. 104 (1972). In the latter context, a defendant has no right to insist on two opportunities to prove his case and rebut the prosecution's. Although there ultimately may be two trials, as when a conviction is reversed on appeal for trial error, that eventuality is largely beyond the defendant's control. A defendant will therefore ordinarily approach a trial on the assumption that it will be his only opportunity to influence the factfinder in his favor. That expectation will presumably result in a maximum dedication of the defendant's resources to the trial, which in turn will engender a significant degree of anxiety during the course of proceedings.
In contrast, as the dissenting judge in the Court of Appeals pointed out, Lydon chose to be tried in a system the defining characteristic of which is that it provides the defendant "two full opportunities to be acquitted on the facts." 698 F.2d, at 11 (Campbell, J., dissenting) (emphasis in original). Unlike a defendant in a traditional trial system, a defendant in Lydon's position knows from the outset of the first-tier proceeding that, at its conclusion, he can demand a chance to convince a second factfinder that he is innocent. This knowledge permits him to adopt in advance a trial strategy based on that opportunity. He can, for example, withhold some of his stronger evidence with the intention of introducing it at the second tier after evaluating the prosecution's entire case; in addition, he can take risks in his presentation, secure in the knowledge that he can avoid any resulting dangers the second time around. Perhaps more importantly, the defendant's realization throughout the first-tier trial that he has an absolute right to a second chance necessarily mitigates the sense of irrevocability that normally attends the factfinding stage of criminal proceedings, from beginning to end. For these reasons, the defendant's prospective knowledge of his entitlement to a second factfinding opportunity substantially diminishes the burden imposed by the first proceeding as well as the significance of a guilty verdict ending that proceeding.
Furthermore, the strategic advantage gained by a defendant who chooses the two-tier system is enhanced by virtue of the fact that the prosecution does not share an equivalent advantage. As the Court notes, the "prosecution has every incentive to put forward its strongest case at the bench trial, because an acquittal will preclude reprosecution of the defendant." Ante, at 311. The Court also notes that "[a]lthough admittedly the Commonwealth at the de novo trial will have the benefit of having seen the defense, the defendant likewise will have had the opportunity to assess the prosecution's case." Ibid. Of course, both of these points could be advanced to justify the retrial of a defendant who has been convicted in a traditional system and who has not appealed — a practice prohibited under the Double Jeopardy Clause. See ante, at 306-307. What distinguishes the Massachusetts system for me, however, is that it permits but does not compel a defendant to secure the advantage of knowing in advance that he, but not the prosecution, may demand a second factfinding opportunity. That advantage substantially reduces the significance of the circumstances surrounding a guilty verdict concluding the first-tier to the point that I conclude that such a verdict does not "terminate" jeopardy.
Of course the features of the two-tier system that I have identified might not be advantageous to every defendant; indeed, the nature of a case or the strength of the government's evidence may be such that those characteristics could prove undesirable or unfair to the defendant. Accordingly, Page 326 I find it significant that those aspects of the Massachusetts two-tier system that depart from a traditional trial are not forced on the defendant. Because the Commonwealth permits a defendant to decide for himself whether to accept the burdens of the two-tier proceeding in exchange for its benefits, I need not decide whether a system that allows no such choice would also survive constitutional scrutiny. Cf. Ludwig v. Massachusetts, 427 U.S., at 632 (STEVENS, J., dissenting). See also Ward v. Village of Monroeville, 409 U.S. 57 (1972).
This conclusion is unaffected by Lydon's claim that earlier Massachusetts cases led him to believe that he could challenge the sufficiency of the evidence presented at the first-tier trial through a motion to dismiss filed at the outset of the second-tier. See Brief for Respondent 55. Cf. post, at 331-332, n. 2 (STEVENS, J., concurring in part and concurring in judgment). Assuming the authoritativeness of those cases and Lydon's reasonable reliance on them, the Commonwealth's failure to provide a promised avenue of relief might amount to a violation of due process. The prospect of such a remedy does not, however, bear on whether the circumstances surrounding a guilty verdict at the end of the first tier "terminated" proceedings for purposes of the Double Jeopardy Clause. Faced with a charge for which he believes the prosecution has constitutionally insufficient evidence, a defendant in Lydon's position can choose the ordinary one-tier system in the expectation that, if his sufficiency claim is sustained, he will never be required to undergo a second trial under Burks. A decision to select the two-tier system instead necessarily achieves the advantages flowing from the knowledge that he can demand a second factfinding opportunity. Even if that choice is made only as a hedge against the possibility that the insufficiency claim will be rejected by every court the defendant believes can entertain it, selection of the two-tier alternative itself clearly diminishes both the strategic and emotional significance of the guilty verdict at the first tier.
For these reasons, I conclude that the guilty verdict rendered at the end of Lydon's bench trial did not, for purposes of the Double Jeopardy Clause, "terminate" one trial and thereby permit a claim that a second trial was barred due to insufficient evidence. Accordingly, I agree that the federal habeas court erred in sustaining Lydon's claim on the merits and therefore join the judgment of the Court.
I agree with JUSTICE O'CONNOR that there is no federal habeas corpus jurisdiction. I continue to believe that Hensley v. Municipal Court, 411 U.S. 345 (1973), was wrongly decided for the reasons indicated by the dissent in that case. But accepting Hensley as the law — as I do — there is no reason to extend it to find that Lydon was in "custody" when he is free on his own recognizance. As JUSTICE O'CONNOR explains, Hensley is best understood as interpreting "custody" to include those cases where a criminal defendant, already convicted and sentenced, would be imprisoned without further state judicial action had not the prison sentence been stayed by the federal court on habeas. The State had "emphatically indicated its determination to put [Hensley] behind bars," id., at 351-352, and would have done so but for a stay by the Federal District Court.
Lydon's petition does not present such a case. Until Lydon is convicted, he is obligated only to appear at trial and to "keep the peace." If the trial court finds that he has defaulted on his recognizance, the court may sentence him pursuant to his first conviction; but Lydon then may seek appellate review, see, e. g., Commonwealth v. Bartlett, 374 Mass. 744, 374 N.E.2d 1203 (1978). It trivializes habeas corpus jurisdiction, historically a protection against governmental oppression, to use it as a remedy against restraints as petty as those to which Lydon is subject.
However, as the Court chooses a different tack, I address the merits as well and join Parts I, II-B, III, and IV of JUSTICE WHITE's opinion.
It is necessary to analyze the character of the substantive claim made by respondent before addressing the more difficult procedural questions. Properly analyzed, respondent's habeas corpus petition raises two distinct constitutional claims: First, whether the entry of a judgment of guilt at the conclusion of his first-tier trial deprived him of liberty without due process of law because the evidence was constitutionally insufficient, and second, whether the second-tier trial, if held before the first question is answered, would violate Lydon's constitutional right not to be twice placed in jeopardy for the same offense.
The answer to the first question is easy. If, as respondent alleged and the District Court found, the Commonwealth's evidence at respondent's first-tier trial was insufficient to support a finding of guilt in the first-tier trial, he was entitled to an acquittal. Such an acquittal would have given respondent his unconditional freedom. Instead, he was found guilty of a crime and sentenced to two years in jail. It is true, of course, that Massachusetts has afforded him a right to have that judgment vacated, but as the Court has demonstrated, that relief does not terminate his custodial status Ante, at 300-302. As a matter of federal constitutional law, he had a right to a judgment of acquittal that would eliminate the restraints on his liberty. The Due Process Clause does not permit a State to deprive a person of liberty based on a finding of guilt beyond reasonable doubt after a proceeding in which it failed to adduce sufficient evidence to persuade any trier of fact of guilt beyond reasonable doubt. Jackson v. Virginia, 443 U.S. 307 (1979). Therefore, respondent's continued custody constitutes a deprivation of liberty without due process of law.
The answer to the second question is more difficult. Petitioners concede and the Court assumes that jeopardy attached at the swearing of the first witness at respondent's first-tier trial. Ante, at 309; see also ante, at 314 (BRENNAN, J., concurring in part and concurring in judgment). The question then becomes whether the Commonwealth now seeks to place respondent in jeopardy a second time. The Court and JUSTICE BRENNAN seem to state that had respondent been acquitted at his first-tier, the Constitution would prohibit the second-tier trial. Ante, at 308; ante, at 318 (BRENNAN, J., concurring in part and concurring in judgment). There is also common ground on the proposition that a judgment of acquittal is a necessary precondition to the success of respondent's double jeopardy claim. The Court says that an acquittal would "terminate" jeopardy; thus a second trial would constitute a new and therefore second and unconstitutional attachment of jeopardy, ante, at 308-309. JUSTICE BRENNAN writes that once a judgment of acquittal is obtained the Constitution prohibits retrial, and frames the question as whether respondent was entitled to such a judgment prior to his second trial, ante, at 317-319.
What makes this case difficult is that the first-tier trial actually ended with a judgment of conviction. Respondent does not rely on that judgment as the bar to the second-tier trial. Instead, the predicate for his double jeopardy claim is a hypothetical judgment that he contends should have been entered at the end of the first trial. I agree with JUSTICE BRENNAN that the Court's use of the concept of "continuing jeopardy" is unhelpful, and that the underlying issue in this case is whether respondent is constitutionally entitled to a judgment of acquittal that could form the predicate for his double jeopardy claim. Ante, at 313-319. To put it another way, until a judgment of acquittal is entered — or until there is an adjudication establishing his right to such a judgment — respondent's double jeopardy claim is premature.
The central procedural question the case presents, therefore, is when, if ever, is respondent entitled to have his first constitutional claim — that he was denied due process as a result of the first-tier trial — adjudicated. This Court, like the Supreme Judicial Court of Massachusetts, answers this question "never." I disagree. If, as I suggest above, respondent's current custody is in violation of the Due Process Clause, then respondent has a due process claim cognizable on federal habeas review under Jackson. If this claim is sustained by the federal habeas court, as it was here, that judgment would provide the predicate for respondent's double jeopardy claim. Such a judgment by the federal habeas court would fall under the rule of Burks v. United States, 437 U.S. 1 (1978). What we said of an appellate court's reversal of a jury verdict there would apply equally to a federal habeas court's judgment that the Commonwealth's evidence at the first-tier trial was insufficient:
"[A]n appellate reversal means that the government's case was so lacking that it should not have been even submitted to the jury. Since we necessarily afford absolute finality to a jury's verdict of acquittal — no matter how erroneous its decision — it is difficult to conceive how society has any greater interest in retrying a defendant when, on review, it is decided as a matter of law that the jury could not properly have returned a verdict of guilty." Id., at 16 (emphasis in original).
See also Tibbs v. Florida, 457 U.S. 31, 41 (1982) ("A verdict of not guilty whether rendered by the jury or directed by the trial judge, absolutely shields the defendant from retrial. A reversal based on insufficiency of the evidence has the same effect because it means that no rational factfinder could have voted to convict the defendant").
In short, if Massachusetts affords respondent no remedy, I believe a federal court must adjudicate respondent's Jackson claim, and, if it is sustained, provide habeas corpus relief in the form of an order that requires the State to enter, nunc pro tunc, the judgment of acquittal to which respondent is constitutionally entitled. If and when such a judgment of acquittal is entered, that judgment would bar a second prosecution for the same offense. Or, if the second prosecution had already been concluded before the judgment of acquittal was entered, any jeopardy associated with the second proceeding would be foreclosed; even if the prosecutor had adduced additional evidence at the second-tier trial, the second judgment could not survive the preclusive effect of the acquittal even though it was belatedly entered.
JUSTICE BRENNAN resists this conclusion in "the unique context of the Massachusetts two-tier trial system" because respondent selected this system and received certain tactical advantages as a result of that decision. Ante, at 324. However, the tactical advantages JUSTICE BRENNAN discusses would be entirely illusory if respondent could be convicted even if the Commonwealth adduced insufficient evidence against him at the first-tier trial. The Massachusetts system is only fair to defendants if it acquits those who deserve acquittal. We do not know whether respondent would have selected this system had he known that he had no right to be acquitted at his first-tier trial even if the Commonwealth's evidence was incapable of persuading any rational trier of fact of his guilt. Surely respondent did not validly waive his right to be acquitted under those circumstances in the sense of intentionally relinquishing a known right, which is what the Constitution requires. See Green v. United States, 355 U.S. 184, 191-192 (1957). See also Burks v. United States, 437 U.S. 1, 17 (1978). Respondent's right to an acquittal if there was a failure of proof at the first-tier trial must be enforced if the quid pro quo which JUSTICE BRENNAN believes validates the Massachusetts system is to be realized. Moreover, if, as petitioners concede and the Court and JUSTICE BRENNAN assume, jeopardy attached when the first witness at respondent's first-tier trial was sworn, double jeopardy would operate to prevent the second-tier trial under JUSTICE BRENNAN's own analysis of the case. As he explains, ante, at 315-318, the Double Jeopardy Clause has been construed to permit jeopardy to "continue" only when there has not been a failure of proof at the Page 332 first trial. See Burks, 437 U.S., at 15-16. Here there has been a failure of proof, and hence, as Burks and JUSTICE BRENNAN explain, no legitimate interest in retrial. Without a valid reason to "continue" jeopardy, the Commonwealth cannot constitutionally subject respondent to continued criminal proceedings. Finally, if the Commonwealth convicted respondent on insufficient evidence at the first-tier trial, that trial was fundamentally unfair and the continued deprivation of respondent's liberty is violative of due process. We have refused to tolerate fundamentally unfair first-tier trials simply because a fair trial will be provided at the second-tier. See Ward v. Village of Monroeville, 409 U.S. 57, 61-62 (1972) (availability of trial de novo does not cure bias of judge at first-tier trial).
This reasoning leads me to what I regard as the most difficult issue in the case — not whether there should be federal review of Lydon's claim, but rather, when that review should take place. In answering that question, it is important to keep in mind the precise issue that the federal court must address. That issue is not, as the Court suggests, whether "Lydon could be retried de novo without any judicial determination of the sufficiency of the evidence at his prior bench trial." Ante, at 303 (footnote omitted). The judge who presided at the first trial did make such a "judicial determination" that the evidence was sufficient. Lydon claims that the determination was erroneous — indeed that the evidence was constitutionally insufficient — but he cannot deny that there was such a judicial determination. What is at issue is whether respondent is entitled to review of the constitutional sufficiency of the prosecutor's evidence under Jackson v. Virginia prior to his second-tier trial.
I join the judgment because I believe it was inappropriate for the District Court to entertain respondent's Jackson claim prior to his second-tier trial. The disruption of orderly state processes attendant to the exercise of federal habeas jurisdiction when state proceedings remain pending weighs strongly, and in my view decisively, against the exercise of jurisdiction.
"This Court has long recognized that in some circumstances considerations of comity and concerns for the orderly administration of criminal justice require a federal court to forgo the exercise of its habeas corpus power." Francis v. Henderson, 425 U.S. 536, 539 (1976). For example, we have held that federal courts should not exercise habeas jurisdiction when the petitioner has failed to comply with state simultaneous-objection rules, because of the weighty state interests underlying enforcement of such rules. See Engle v. Isaac, 456 U.S. 107 (1982); Wainwright v. Sykes, 433 U.S. 72 (1977).
One of the weightiest of state interests is that favoring speedy, efficient, and uninterrupted disposition of criminal cases. Because of this critical state interest, we have held that federal courts should abstain from exercising their jurisdiction when the effect thereof would be to disrupt ongoing state proceedings. See, e. g., Hicks v. Miranda, 422 U.S. 332, 349 (1975); Huffman v. Pursue, Ltd., 420 U.S. 592, 599-601 (1975); Perez v. Ledesma, 401 U.S. 82, 84-85 (1971); Younger v. Harris, 401 U.S. 37, 41-45 (1971).
Similarly, the statutory exhaustion requirement found in the habeas statute, 28 U.S.C. § 2254, reflects a recognition that federal habeas courts should not disrupt ongoing state proceedings. See Rose v. Lundy, 455 U.S. 509, 518 (1982). Indeed, in our leading case concerning the propriety of pretrial federal habeas intervention under the exhaustion doctrine, we cautioned that such review would be inappropriate when it threatens to disrupt pending state proceedings and orderly state processes. See Braden v. 30th Judicial Circuit Court of Kentucky, 410 U.S. 484, 490-493 (1973). Thus, the habeas statute itself reflects this concern with disrupting ongoing state proceedings.
I am not suggesting that respondent's double jeopardy claim has not been exhausted; I agree that it has been for the reasons stated in Part II-B of the opinion of the Court. However, while that claim has been exhausted, it would nevertheless be meritless unless the antecedent Jackson claim may also be entertained by the federal habeas court. As to that claim it is true that in a technical sense respondent may well have no state Page 334 remedy to exhaust inasmuch as the Massachusetts courts have indicated that they will not review respondent's Jackson claim even after his second-tier trial. See ante, at 322-323, n. 6 (BRENNAN, J., concurring in part and concurring in judgment). However, even if there has been exhaustion in a technical sense here, the more fundamental policies underlying the exhaustion requirement may be jeopardized if a habeas petition is entertained while state proceedings remain pending. After all, exhaustion was originally a judge-made rule designed not as a technical doctrine but rather to prevent premature and unjustified interference in state proceedings. See, e. g., Ex parte Hawk, 321 U.S. 114, 116-118 (1944) (per curiam); United States ex rel. Kennedy v. Tyler, 269 U.S. 13, 17-19 (1925); Davis v. Burke, 179 U.S. 399, 402-403 (1900); Ex parte Royall, 117 U.S. 241, 251-252 (1886).
The state interest against disruption of ongoing proceedings is squarely implicated by the exercise of federal habeas jurisdiction over this case. Respondent was convicted at his first-tier bench trial on November 20, 1979, and his second-tier jury trial was originally set for November 29. That trial has been delayed for over four years. While some of that delay has been attributable to litigation in the state courts, over three years' worth of delay is attributable to federal habeas review.
This case was pending approximately seven months in the District Court, and in the Court of Appeals about another seven months. By this observation I intend no criticism of these courts. If anything, both courts disposed of the case with more than reasonable promptness. Rather, I make this observation to demonstrate the inevitable delay whenever federal habeas review is commenced, even if the case is adjudicated with commendable dispatch.
If we were to uphold the exercise of federal habeas jurisdiction here, similar delays could become routine in Massachusetts. Already there are some 14,000 cases a year taken to the second-tier jury trial. In virtually all of these cases, the defendant could seek federal habeas review at the conclusion of the first trial, claiming that the evidence used to convict him was insufficient. Defendants have every incentive to seek habeas review, not only to delay eventual punishment, but to obtain leverage in plea negotiations. The speed and efficiency of the process would quickly be eroded if collateral litigation intervened between the first and second trials. The wholesale disruption of pending proceedings that would occur if federal habeas review were available between the first and second trials to every defendant who thought the evidence of his guilt was insufficient counsels strongly against the exercise of such jurisdiction. The state process should be permitted to proceed in an uninterrupted fashion before federal habeas review comes into play.
I have no doubt that if we approved the exercise of habeas jurisdiction in this case, the district judges in Massachusetts would attempt to minimize disruption by adjudicating habeas cases as quickly as possible. Nevertheless, the quality of justice in such a harried process is bound to suffer. Moreover, the district judges in Massachusetts, as elsewhere, have enough burdens with which they must cope without the additional time pressure created by "interlocutory" habeas cases such as this one.
Respondent and the Court of Appeals suggest that habeas review could be limited to cases in which the petitioner could make a strong initial showing of a likely constitutional violation. Nevertheless, every defendant could attempt to make such a showing in the few days between the first- and second-tier trials. Such hurry-up litigation will burden prosecutors and courts, reduce the quality of justice, and surely prove impractical (it will certainly take more than a few days just to obtain the record and transcribe the recording of the first-tier trial), forcing the state system to delay until the federal case can be adjudicated.
The postponement of review in this case would not render petitioner's double jeopardy claim entirely nugatory. First, if respondent's claim is meritorious, under my view, he would ultimately obtain relief from his conviction through federal habeas review after state proceedings are complete. Moreover, if his claim is meritorious, respondent will likely be acquitted at his second-tier trial precisely because of the insufficiency of the Commonwealth's evidence. It is true, of course, that the prosecutor may supply proof of an element of the offense that was omitted in the first trial. It is reasonable to assume, however, that in most of the relatively simple misdemeanor prosecutions that employ this procedure, the same evidence will again be offered and the same issue will again be presented to the second judge as to the first. The likelihood that the substance of respondent's claim will be heard and vindicated at his impending trial argues all the more strongly against federal intervention at this point in the proceedings.
In this case the District Court's findings indicate that the essential problem with the Commonwealth's case is that respondent was charged with the wrong offense. That problem cannot be remedied simply by adducing additional evidence at the second-tier trial.
Second, if my view were to prevail, state prosecutors would be aware that the sufficiency of the evidence at the first-tier trial would eventually be reviewed, and they would therefore have a greater incentive to adduce sufficient evidence at that trial. Thus, the ultimate availability of federal collateral review would reduce the likelihood of a constitutional violation.
Finally, as the Court explains, ante, at 310-312, the Massachusetts two-tier trial system is not an especially harsh one. By voluntarily electing that procedure, the defendant has accepted the risk of two trials when he could insist upon only one. While this election cannot justify a refusal to provide any remedy for a constitutional violation, it does indicate that the enforcement of the exhaustion requirement in this case would not place upon respondent an entirely unavoidable obligation to endure two trials.
On balance I think the principles of comity that underlie the exhaustion and abstention doctrines make the exercise of federal habeas jurisdiction in this case premature. The state interest in avoiding wholesale disruption of its criminal process requires a federal habeas court to postpone the exercise of its jurisdiction over this case until after the second-tier trial has been completed. I would hold that in order to assert his constitutional claims, respondent must first take advantage of the opportunity the State provides him for an acquittal in the second trial. If he is convicted in that proceeding, I would hold that a federal court may then review the record of the first trial to determine whether he was constitutionally entitled to an acquittal. If the record should then support the claim that respondent has made, I would conclude that he is entitled to release even if the State adduced enough additional evidence at the second-tier trial to support a conviction. Accordingly, I concur in Parts I and II of the Court's opinion and in the judgment.
I agree that the judgment of the Court of Appeals should be reversed. Unlike the Court, however, I conclude that the District Court lacked jurisdiction to hear respondent Lydon's habeas petition at this stage in the ongoing state-court proceeding.
The Court suggests that federal habeas jurisdiction exists whenever (i) a state defendant is subject to minimal legal restraints on his freedom and (ii) the defendant has exhausted state avenues of relief with respect to the particular federal claim brought to the habeas court. Then, recognizing that its unadorned test might greatly expand federal habeas jurisdiction, the Court, ante, at 302, emphasizes "the unique nature of the double jeopardy right." In my view the Court first unnecessarily expands the holding in Hensley v. Municipal Court, 411 U.S. 345 (1973), and then limits the damage by restricting its exhaustion analysis to double jeopardy claims. I would prefer to search for a more principled understanding of the statutory term "custody."
Under Massachusetts law, as I read it, Lydon is no longer in custody "pursuant" to the judgment entered at his first trial. Lydon has invoked his right to a second trial and appeared at the second proceeding. Under Massachusetts law, therefore, the results of the first trial — together with any incidental "custody" imposed in consequence of that trial — have already been eliminated. The restraints on Lydon's freedom now derive not from the prior conviction, but from the fact that a new criminal proceeding is in progress. Every state defendant who fails to attend a criminal trial risks punitive sanctions not dissimilar to those to which Lydon is currently exposed.
Federal habeas jurisdiction plainly does not attach merely because a state criminal defendant, whose freedom to come and go as he pleases is limited in some way in connection with a criminal proceeding, has exhausted state interlocutory review of a particular federal claim. Federal habeas jurisdiction is absent because "custody" in connection with an ongoing trial is usually not "in violation of the Constitution or laws or treaties of the United States," 28 U.S.C. § 2241(c)(3), 2254(a), even when the proceedings themselves or the underlying charge are constitutionally defective. Most constitutional rights exist to protect a criminal defendant from conviction — not from the process of trial itself.
In this regard, however, I agree with the Court that double jeopardy is different. Here, custody incident to a trial may violate the Constitution because the trial itself, regardless of its outcome, is unconstitutional. For this reason I agree that a prisoner who is incarcerated in connection with a criminal proceeding is "in custody in violation of the Constitution," 28 U.S.C. § 2254(a), when the proceeding violates his double jeopardy rights. Cf. Arizona v. Washington, 434 U.S. 497 (1978). But I do not agree that the minor restraints on Lydon's freedom, incurred in connection with an ongoing state trial, satisfy the jurisdictional requirements of the habeas statute. Nor do I believe that Hensley dictates a different result.
In Hensley the Court made it quite clear that a relaxed definition of "custody" was accepted only because incarceration was imminent and, absent federal intervention, inevitable. The habeas petitioner in Hensley had exhausted "all available state court opportunities to have [his] conviction set aside," 411 U.S., at 353; see also id., at 346, 347, and n. 4, 351, 352, not merely all available court opportunities to review the particular claim in question. Hensley emphasized that the typical restrictions on freedom attending a release on personal recognizance would not, standing alone, constitute "custody" within the meaning of the habeas statute. Such restraints amount to "custody" only when state judicial proceedings have been completed and incarceration has become a purely executory decision. Hensley accepted a liberal definition of "custody" only in conjunction with an unusual requirement of absolute exhaustion — exhaustion not of the particular claim in question, cf. 28 U.S.C. § 2254(b), but of all possible state avenues of relief from the conviction.
My reading of Hensley thus leads me to conclude that a state criminal defendant should be considered "in custody pursuant to the judgment of a State court," 28 U.S.C. § 2254(a), only when he is under physical restraint, cf. Arizona v. Washington, supra, or under a legal restraint that can be converted into physical restraint without a further judicial hearing. The latter situation will normally arise only when state judicial proceedings (as distinguished from particular claims raised in those proceedings) have been entirely exhausted.
Even if the habeas petitioner is in physical custody, it may well be appropriate for a federal court to abstain from deciding the petition until state-court proceedings have been completed.
Lydon's condition clearly does not meet the Hensley test as I understand it. Lydon has not come close to exhausting state opportunities to have the conviction set aside. Lydon cannot be incarcerated without a further judicial hearing. His position is thus functionally indistinguishable from that of a defendant pressing an interlocutory appeal. One claim may have been exhausted, but others have not. In these circumstances, incarceration is far from inevitable, and the minor constraints that attend a release on personal recognizance are much less significant. If Massachusetts stood ready to incarcerate Lydon on the basis of the conviction at the first trial my view of the case would be different.
The Court makes clear, ante, at 302-303, its view that double jeopardy claims are "unique" for federal habeas purposes. This might be sufficient reason to bring such a claim within Hensley's rationale even when only the specific claim has been exhausted. Cf. Abney v. United States, 431 U.S. 651 (1977); Arizona v. Washington, supra; Braden v. 30th Judicial Circuit Court of Kentucky, 410 U.S. 484 (1973). For my part, I would prefer to avoid relaxing Hensley's clear holding that the minimal constraints of a release on personal recognizance constitute "custody" only when the State stands ready to incarcerate the habeas petitioner without further judicial hearing. A special purpose jurisdictional exception for double jeopardy allegations seems inadvisable simply because the habeas statute contains no license for such an exception. "Custody" is the touchstone relied on by § 2254; of all the possible unconstitutional infringements on personal freedom, only unlawful "custody" has been identified as providing a sufficient basis for federal intervention. I would therefore hold that a state criminal defendant is not "in custody pursuant to the judgment of a State court" while he remains free from physical restraint and the State remains unable to impose such restraint without a further judicial hearing.