JUSTICE THOMAS, concurring.

The Constitution charges federal judges with deciding cases and controversies, not with running state prisons. Yet, too frequently, federal district courts in the name of the Constitution effect wholesale takeovers of state correctional facilities and run them by judicial decree. This case is a textbook example. Dissatisfied with the quality of the law libraries and the legal assistance at Arizona's correctional institutions, the District Court imposed a statewide decree on the Arizona Department of Corrections (ADOC), dictating in excruciatingly minute detail a program to assist inmates in the filing of lawsuits — right down to permissible noise levels in library reading rooms. Such gross overreaching by a federal district court simply cannot be tolerated in our federal system. Principles of federalism and separation of powers dictate that exclusive responsibility for administering state prisons resides with the State and its officials.

Of course, prison officials must maintain their facilities consistent with the restrictions and obligations imposed by the Constitution. In Bounds v. Smith, 430 U.S. 817 (1977), we recognized as part of the State's constitutional obligations a duty to provide prison inmates with law libraries or other legal assistance at state expense, an obligation we described as part of a loosely defined "right of access to the courts" enjoyed by prisoners. While the Constitution may guarantee state inmates an opportunity to bring suit to vindicate their federal constitutional rights, I find no basis in the Constitution — and Bounds cited none — for the right to have the government finance the endeavor.

I join the majority opinion because it places sensible and much-needed limitations on the seemingly limitless right to assistance created in Bounds and because it clarifies the scope of the federal courts' authority to subject state prisons to remedial decrees. I write separately to make clear my doubts about the validity of Bounds and to reiterate my observation in Missouri v. Jenkins, 515 U.S. 70 (1995), that the federal judiciary has for the last half century been exercising "equitable" powers and issuing structural decrees entirely out of line with its constitutional mandate.

I A

This case is not about a right of "access to the courts." There is no proof that Arizona has prevented even a single inmate from filing a civil rights lawsuit or submitting a petition for a writ of habeas corpus. Instead, this case is about the extent to which the Constitution requires a State to finance or otherwise assist a prisoner's efforts to bring suit against the State and its officials.

In Bounds v. Smith, supra, we recognized for the first time a "fundamental constitutional right" of all inmates to have the State "assist [them] in the preparation and filing of meaningful legal papers." Id., at 828. We were not explicit as to the forms the State's assistance must take, but we did hold that, at a minimum, States must furnish prisoners "with adequate law libraries or adequate assistance from persons trained in the law." Ibid. Although our cases prior to Bounds occasionally referenced a constitutional right of access to the courts, we had never before recognized a freestanding constitutional right that requires the States to "shoulder affirmative obligations," id., at 824, in order to "insure that inmate access to the courts is adequate, effective, and meaningful," id., at 822.

Recognition of such broad and novel principles of constitutional law are rare enough under our system of law that I would have expected the Bounds Court to explain at length the constitutional basis for the right to state-provided legal materials and legal assistance. But the majority opinion in Bounds failed to identify a single provision of the Constitution to support the right created in that case, a fact that did not go unnoticed in strong dissents by Chief Justice Burger and then-JUSTICE REHNQUIST. See id., at 833-834 (Opinion of Burger, C. J.,) ("The Court leaves us unenlightened as to the source of the `right of access to the courts' which it perceives or of the requirement that States `foot the bill' for assuring such access for prisoners who want to act as legal researchers and brief writers"); id., at 840 (Opinion of REHNQUIST, J.) ("[T]he `fundamental constitutional right of access to the courts' which the Court announces today is created virtually out of whole cloth with little or no reference to the Constitution from which it is supposed to be derived"). The dissents' calls for an explanation as to which provision of the Constitution guarantees prisoners a right to consult a law library or a legal assistant, however, went unanswered. This is perhaps not surprising: Just three years before Bounds was decided we admitted that the "[t]he precise rationale" for many of the "access to the courts" cases on which Bounds relied had "never been explicitly stated," and that no Clause that had thus far been advanced "by itself provides an entirely satisfactory basis for the result reached." Ross v. Moffitt, 417 U.S. 600, 608-609 (1974).

The weakness in the Court's constitutional analysis in Bounds is punctuated by our inability, in the 20 years since, to agree upon the constitutional source of the supposed right. We have described the right articulated in Bounds as a "consequence" of due process, Murray v. Giarratano, 492 U.S. 1, 11, n. 6 (1989) (plurality opinion) (citing Procunier v. Martinez, 416 U.S. 396, 419 (1974)), as an "aspect" of equal protection, 492 U.S., at 11, n. 6 (citation omitted), or as an "equal protection guarantee," Pennsylvania v. Finley, 481 U.S. 551, 557 (1987). In no instance, however, have we engaged in rigorous constitutional analysis of the basis for the asserted right. Thus, even as we endeavor to address the question presented in this case — whether the District Court's order "exceeds the constitutional requirements set forth in Bounds," Pet. for Cert. i — we do so without knowing which Amendment to the Constitution governs our inquiry.

It goes without saying that we ordinarily require more exactitude when evaluating asserted constitutional rights. "As a general matter, the Court has always been reluctant" to extend constitutional protection to "unchartered area[s]," where the "guideposts for responsible decisionmaking . . . are scarce and open-ended." Collins v. Harker Heights, 503 U.S. 115, 125 (1992). It is a bedrock principle of judicial restraint that a right be lodged firmly in the text or tradition of a specific constitutional provision before we will recognize it as fundamental. Strict adherence to this approach is essential if we are to fulfill our constitutionally assigned role of giving full effect to the mandate of the Framers without infusing the constitutional fabric with our own political views.

B

In lieu of constitutional text, history, or tradition, Bounds turned primarily to precedent in recognizing the right to state assistance in the researching and filing of prisoner claims. Our cases, however, had never recognized a right of the kind articulated in Bounds, and, in my opinion, could not reasonably have been read to support such a right. Prior to Bounds, two lines of cases dominated our so-called "access to the courts" jurisprudence. One of these lines, rooted largely in principles of equal protection, invalidated state filing and transcript fees and imposed limited affirmative obligations on the States to ensure that their criminal procedures did not discriminate on the basis of poverty. These cases recognized a right to equal access, and any affirmative obligations imposed ( e.g., a free transcript or counsel on a first appeal as of right) were strictly limited to ensuring equality of access, not access in its own right. In a second line of cases, we invalidated state prison regulations that restricted or effectively prohibited inmates from filing habeas corpus petitions or civil rights lawsuits in federal court to vindicate federally protected rights. While the cases in this line did guarantee a certain amount of access to the federal courts, they imposed no affirmative obligations on the States to facilitate access, and held only that States may not "abridge or impair" prisoners' efforts to petition a federal court for vindication of federal rights. Ex parte Hull, 312 U.S. 546, 549 (1941). Without pausing to consider either the reasoning behind, or the constitutional basis for, each of these independent lines of case law, the Court in Bounds engaged in a loose and selective reading of our precedents as it created a freestanding and novel right to state-supported legal assistance. Despite the Court's purported reliance on prior cases, Bounds in fact represented a major departure both from precedent and historical practice.

1

In a series of cases beginning with Griffin v. Illinois, 351 U.S. 12 (1956), the Court invalidated state rules that required indigent criminal defendants to pay for trial transcripts or to pay other fees necessary to have their appeals or habeas corpus petitions heard. According to the Bounds Court, these decisions "struck down restrictions and required remedial measures to insure that inmate access to the courts is adequate, effective, and meaningful." 430 U.S., at 822. This is inaccurate. Notwithstanding the suggestion in Bounds, our transcript and fee cases did not establish a freestanding right of access to the courts, meaningful or otherwise.

In Griffin, for instance, we invalidated an Illinois rule that charged criminal defendants a fee for a trial transcript necessary to secure full direct appellate review of a criminal conviction. See 351 U.S., at 13-14; id., at 22 (Frankfurter, J., concurring in judgment). See also Ross v. Moffitt, supra at 605-606. Though we held the fee to be unconstitutional, our decision did not turn on the effectiveness or adequacy of the access afforded to criminal defendants generally. We were quite explicit in reaffirming the century-old principle that "a State is not required by the Federal Constitution to provide appellate courts or a right to appellate review at all." Griffin, supra, at 18 (emphasis added) (citing McKane v. Durston, 153 U.S. 684, 687-688 (1894)). Indeed, the Court in Griffin was unanimous on this point. See 351 U.S., at 21 (Frankfurter, J., concurring in judgment) ("[I]t is now settled that due process of law does not require a State to afford review of criminal judgments"); id., at 27 (Burton, J., dissenting) ("Illinois, as the majority admit, could thus deny an appeal altogether in a criminal case without denying due process of law"); id., at 36 (Harlan, J., dissenting) ("The majority of the Court concedes that the Fourteenth Amendment does not require the States to provide for any kind of appellate review"). In light of the Griffin Court's unanimous pronouncement that a State is not constitutionally required to provide any court access to criminals who wish to challenge their convictions, the Bounds Court's description of Griffin as ensuring "`adequate and effective appellate review,'" 430 U.S., at 822 (quoting Griffin, supra, at 20), is unsustainable.

Instead, Griffin rested on the quite different principle that, while a State is not obliged to provide appeals in criminal cases, the review a State chooses to afford must not be administered in a way that excludes indigents from the appellate process solely on account of their poverty. There is no mistaking the principle that motivated Griffin:

"It is true that a State is not required by the Federal Constitution to provide appellate courts or a right to appellate review at all. But that is not to say that a State that does grant appellate review can do so in a way that discriminates against some convicted defendants on account of their poverty. . . . [A]t all stages of the proceedings the Due Process and Equal Protection Clauses protect [indigent persons] from invidious discriminations. . . .

". . . There can be no equal justice where the kind of trial a man gets depends on the amount of money he has. Destitute defendants must be afforded as adequate appellate review as defendants who have money enough to buy transcripts." 351 U.S., at 18-19 (plurality opinion) (citation omitted).

Justice Frankfurter, who provided the fifth vote for the majority, confirmed in a separate writing that it was invidious discrimination, and not the denial of adequate, effective, or meaningful access to the courts, that rendered the Illinois regulation unconstitutional: "[W]hen a State deems it wise and just that convictions be susceptible to review by an appellate court, it cannot by force of its exactions draw a line which precludes convicted indigent persons . . . from securing such a review. . . ." Id., at 23 (opinion concurring in judgment). Thus, contrary to the characterization in Bounds, Griffin stands not for the proposition that all inmates are entitled to adequate appellate review of their criminal convictions, but for the more modest rule that, if the State chooses to afford appellate review, it "can no more discriminate on account of poverty than on account of religion, race, or color." Griffin, supra, at 17 (plurality opinion).

If we left any doubt as to the basis of our decision in Griffin, we eliminated it two decades later in Douglas v. California, 372 U.S. 353 (1963), where we held for the first time that States must provide assistance of counsel on a first appeal as of right for all indigent defendants. Like Griffin, Douglas turned not on a right of access per se, but rather on the right not to be denied, on the basis of poverty, access afforded to others. We did not say in Douglas that indigents have a right to a "meaningful appeal" that could not be realized absent appointed counsel. Cf. Bounds, 430 U.S., at 823. What we did say is that, in the absence of state-provided counsel, "[t]here is lacking that equality demanded by the Fourteenth Amendment where the rich man, who appeals as of right, enjoys the benefit of counse[l] . . . while the indigent . . . is forced to shift for himself." Douglas, supra, at 357-358. Just as in Griffin, where "we held that a State may not grant appellate review in such a way as to discriminate against some convicted defendants on account of their poverty," Douglas, 372 U.S., at 355, the evil motivating our decision in Douglas was "discrimination against the indigent," ibid.

Our transcript and fee cases were, therefore, limited holdings rooted in principles of equal protection. In Bounds, these cases were recharacterized almost beyond recognition, as the Court created a new and different right on behalf of prisoners — a right to have the State pay for law libraries or other forms of legal assistance without regard to the equality of access. Only by divorcing our prior holdings from their reasoning, and by elevating dicta over constitutional principle, was the Court able to reach such a result.

The unjustified transformation of the right to nondiscriminatory access to the courts into the broader, untethered right to legal assistance generally would be reason enough for me to conclude that Bounds was wrongly decided. However, even assuming that Bounds properly relied upon the Griffin line of cases for the proposition for which those cases actually stood, the Bounds Court failed to address a significant intervening development in our jurisprudence: the fact that the equal protection theory underlying Griffin and its progeny had largely been abandoned prior to Bounds. The provisions invalidated in our transcript and fee cases were all facially neutral administrative regulations that had a disparate impact on the poor; there is no indication in any of those cases that the State imposed the challenged fee with the purpose of deliberately discriminating against indigent defendants. See, e.g., Douglas, supra, at 361 (Harlan, J., dissenting) (criticizing the Court for invalidating a state law "of general applicability" solely because it "may affect the poor more harshly than it does the rich"). In the years between Douglas and Bounds, however, we rejected a disparate impact theory of the Equal Protection Clause. That the doctrinal basis for Griffin and its progeny has largely been undermined — and in fact had been before Bounds was decided — confirms the invalidity of the right to law libraries and legal assistance created in Bounds.

We first cast doubt on the proposition that a facially neutral law violates the Equal Protection Clause solely because it has a disparate impact on the poor in San Antonio Independent School Dist. v. Rodriguez, 411 U.S. 1 (1973). In Rodriguez, the respondents challenged Texas' traditional system of financing public education under the Equal Protection Clause on the ground that, under that system, "some poorer people receive less expensive educations than other more affluent people." Id., at 19. In rejecting the claim that this sort of disparate impact amounted to unconstitutional discrimination, we declined the respondents' invitation to extend the rationale of Griffin, Douglas, and similar cases. We explained that, under those cases, unless a group claiming discrimination on the basis of poverty can show that it is " completely unable to pay for some desired benefit, and as a consequence, . . . sustained an absolute deprivation of a meaningful opportunity to enjoy that benefit," 411 U.S., at 20 (emphasis added), strict scrutiny of a classification based on wealth does not apply. Because the respondents in Rodriguez had not shown that "the children in districts having relatively low assessable property values are receiving no public education," but rather claimed only that "they are receiving a poorer quality education than that available to children in districts having more assessable wealth," id., at 23 (emphasis added), we held that the "Texas system does not operate to the peculiar disadvantage of any suspect class," id., at 28. After Rodriguez, it was clear that "wealth discrimination alone [does not] provid[e] an adequate basis for invoking strict scrutiny," id., at 29, and that, "at least where wealth is involved, the Equal Protection Clause does not require absolute equality or precisely equal advantages," id., at 24. See also Kadrmas v. Dickinson Public Schools, 487 U.S. 450, 458 (1988); Harris v. McRae, 448 U.S. 297, 322-323 (1980); Maher v. Roe, 432 U.S. 464, 470-471 (1977).

We rejected a disparate impact theory of the Equal Protection Clause altogether in Washington v. Davis, 426 U.S. 229, 239 (1976), decided just one Term before Bounds. There we flatly rejected the idea that "a law, neutral on its face and serving ends otherwise within the power of government to pursue, is invalid under the Equal Protection Clause simply because it may affect a greater proportion of one race than of another." 426 U.S., at 242. We held that, absent proof of discriminatory purpose, a law or official act does not violate the Constitution " solely because it has a . . . disproportionate impact." Id., at 239 (emphasis in original). See also id., at 240 (acknowledging "the basic equal protection principle that the invidious quality of a law claimed to be racially discriminatory must ultimately be traced to a racially discriminatory purpose"). At bottom, Davis was a recognition of "the settled rule that the Fourteenth Amendment guarantees equal laws, not equal results." Personnel Administrator of Mass. v. Feeney, 442 U.S. 256, 273 (1979).

The Davis Court was motivated in no small part by the potentially radical implications of the Griffin/Douglas rationale. As Justice Harlan recognized in Douglas: "[E]very financial exaction which the State imposes on a uniform basis is more easily satisfied by the well-to-do than by the indigent." 372 U.S., at 361 (dissenting opinion). Under a disparate impact theory, Justice Harlan argued, regulatory measures always considered to be constitutionally valid, such as sales taxes, state university tuition, and criminal penalties, would have to be struck down. See id., at 361-362. Echoing Justice Harlan, we rejected in Davis the disparate impact approach in part because of the recognition that "[a] rule that a statute designed to serve neutral ends is nevertheless invalid, absent compelling justification, if in practice it benefits or burdens one race more than another would be far reaching and would raise serious questions about, and perhaps invalidate, a whole range of tax, welfare, public service, regulatory, and licensing statutes that may be more burdensome to the poor and to the average black than to the more affluent white." 426 U.S., at 248. See also id., at 248, n. 14.

Given the unsettling ramifications of a disparate impact theory, it is not surprising that we eventually reached the point where we could no longer extend the reasoning of Griffin and Douglas. For instance, in Ross v. Moffitt, 417 U.S. 600 (1974), decided just three years before Bounds, we declined to extend Douglas to require States to provide indigents with counsel in discretionary state appeals or in seeking discretionary review in this Court. We explained in Ross that "[t]he Fourteenth Amendment `does not require absolute equality or precisely equal advantages,'" 417 U.S., at 612 (quoting Rodriguez, 411 U.S., at 24), and that it "does [not] require the State to `equalize economic conditions,'" 417 U.S., at 612 (quoting Griffin, 351 U.S., at 23 (Frankfurter, J., concurring in judgment)). We again declined to extend Douglas in Pennsylvania v. Finley, 481 U.S., at 555, where we rejected a claim that the Constitution requires the States to provide counsel in state postconviction proceedings. And we found Ross and Finley controlling in Murray v. Giarratano, 492 U.S. 1 (1989), where we held that defendants sentenced to death, like all other defendants, have no right to state-appointed counsel in state collateral proceedings. See also United States v. MacCollom, 426 U.S. 317 (1976) (federal habeas statute permitting district judge to deny free transcript to indigent petitioner raising frivolous claim does not violate the Constitution).

In sum, the Bounds Court's reliance on our transcript and fee cases was misplaced in two significant respects. First, those cases did not stand for the proposition for which Bounds cited them: They were about equal access, not access per se. Second, the constitutional basis for Griffin and its progeny had been seriously undermined in the years preceding Bounds. Thus, even to the extent that Bounds intended to rely on those cases for the propositions for which they actually stood, their underlying rationale had been largely discredited. These cases, rooted in largely obsolete theories of equal protection, do not support the right to law libraries and legal assistance recognized in Bounds. Our repeated holdings declining to extend these decisions only confirm this conclusion.

2

The Bounds Court relied on a second line of cases in announcing the right to state-financed law libraries or legal assistance for prisoners. These cases, beginning with our decision in Ex parte Hull, prevent the States from imposing arbitrary obstacles to attempts by prisoners to file claims asserting federal constitutional rights. Although this line deals with access in its own right, and not equal access as in Griffin and Douglas, these cases do not impose any affirmative obligations on the States to improve the prisoners' chances of success.

Bounds identified Ex parte Hull as the first case to "recogniz[e]" a "constitutional right of access to the courts." 430 U.S., at 821-822. In Ex parte Hull, we considered a prison regulation that required prisoners to submit their habeas corpus petitions to a prison administrator before filing them with the court. Only if the administrator determined that a petition was "`properly drawn'" could the prisoner submit it in a federal court. 312 U.S., at 548-549 (quoting regulation). We invalidated the regulation, but the right we acknowledged in doing so bears no resemblance to the right generated in Bounds.

Our reasoning in Ex parte Hull consists of a straightforward, and rather limited, principle:

"[T]he state and its officers may not abridge or impair petitioner's right to apply to a federal court for a writ of habeas corpus. Whether a petition for writ of habeas corpus addressed to a federal court is properly drawn and what allegations it must contain are questions for that court alone to determine." 319 U.S., at 549.

The "right of access" to the courts articulated in Ex parte Hull thus imposed no affirmative obligations on the States; we stated only that a State may not "abridge or impair" a prisoner's ability to file a habeas petition in federal court. Ex parte Hull thus provides an extraordinarily weak starting point for concluding that the Constitution requires States to fund and otherwise assist prisoner legal research by providing law libraries or legal assistance.

Two subsequent decisions of this Court worked a moderate expansion of Ex parte Hull. The first, Johnson v. Avery, 393 U.S. 483 (1969), invalidated a Tennessee prison regulation that prohibited inmates from advising or assisting one another in the preparation of habeas corpus petitions. In striking down the regulation, the Court twice quoted Ex parte Hull's holding that a State may not "abridge or impair" a petitioner's efforts to file a petition for a writ of habeas corpus. See 393 U.S., at 486-487, 488. In contrast to Ex parte Hull, however, Johnson focused not on the respective institutional roles of state prisons and the federal courts but on "the fundamental importance of the writ of habeas corpus in our constitutional scheme." 393 U.S., at 485. Still, the Court did not hold that the Constitution places an affirmative obligation on the States to facilitate the filing of habeas petitions. The Court held only that a State may not "den[y] or obstruc[t]" a prisoner's ability to file a habeas petition. Ibid. We extended the holding of Johnson in Wolff v. McDonnell, 418 U.S. 539 (1974), where we struck down a similar regulation that prevented inmates from assisting one another in the preparation of civil rights complaints. We held that the "right of access to the courts, upon which Avery was premised, is founded in the Due Process Clause and assures that no person will be denied the opportunity to present to the judiciary allegations concerning violations of fundamental constitutional rights." Id., at 579. Again, the right was framed exclusively in the negative. See ibid. (opportunity to file a civil rights action may not be "denied"). Thus, prior to Bounds, "if a prisoner incarcerated pursuant to a final judgment of conviction [was] not prevented from physical access to the federal courts in order that he may file therein petitions for relief which Congress has authorized those courts to grant, he ha[d] been accorded the only constitutional right of access to the courts that our cases ha[d] articulated in a reasoned way." Bounds, 430 U.S., at 839-840 (REHNQUIST, J., dissenting) (citing Ex parte Hull).

C

That Ex parte Hull, Johnson, and Wolff were decided on different constitutional grounds from Griffin and Douglas is clear enough. According to Bounds, however, "[e]ssentially the same standards of access were applied" in all of these cases. 430 U.S., at 823. This observation was wrong, but the equation of these two lines of cases allowed the Bounds Court to preserve the "affirmative obligations" element of the equal access cases, the rationale of which had largely been undermined prior to Bounds, by linking it with Ex parte Hull, which had not been undermined by later cases but which imposed no affirmative obligations. In the process, Bounds forged a right with no basis in precedent or constitutional text: a right to have the State "shoulder affirmative obligations" in the form of law libraries or legal assistance to ensure that prisoners can file meaningful lawsuits. By detaching Griffin's right to equal access and Ex parte Hull's right to physical access from the reasoning on which each of these rights was based, the Bounds Court created a virtually limitless right. And though the right was framed in terms of law libraries and legal assistance in that case, the reasoning is much broader, and this Court should have been prepared under the Bounds rationale to require the appointment of capable state-financed counsel for any inmate who wishes to file a lawsuit. See Bounds, supra, at 841 (REHNQUIST, J., dissenting) (observing that "the logical destination of the Court's reasoning" in Bounds is "lawyers appointed at the expense of the State"). See also ante, at 354. We have not, however, extended Bounds to its logical conclusion. And though we have not overruled Bounds, we have undoubtedly repudiated its reasoning in our consistent rejection of the proposition that the States must provide counsel beyond the trial and first appeal as of right. See Ross, 417 U.S., at 612; Finley, 481 U.S., at 555; Giarratano, 492 U.S., at 3-4 (plurality opinion).

In the end, I agree that the Constitution affords prisoners what can be termed a right of access to the courts. That right, rooted in the Due Process Clause and the principle articulated in Ex parte Hull, is a right not to be arbitrarily prevented from lodging a claimed violation of a federal right in a federal court. The State, however, is not constitutionally required to finance or otherwise assist the prisoner's efforts, either through law libraries or other legal assistance. Whether to expend state resources to facilitate prisoner lawsuits is a question of policy and one that the Constitution leaves to the discretion of the States.

There is no basis in history or tradition for the proposition that the State's constitutional obligation is any broader. Although the historical record is relatively thin, those who have explored the development of state-sponsored legal assistance for prisoners agree that, until very recently, law libraries in prisons were "nearly nonexistent." A. Flores, Werner's Manual for Prison Law Libraries 1 (2d ed. 1990). Prior to Bounds, prison library collections (to the extent prisons had libraries) commonly reflected the correctional goals that a State wished to advance, whether religious, educational, or rehabilitative. Although some institutions may have begun to acquire a minimal collection of legal materials in the early part of this century, lawbooks generally were not included in prison libraries prior to the 1950's. See W. Coyle, Libraries in Prisons 54-55 (1987). The exclusion of lawbooks was consistent with the recommendation of the American Prison Association, which advised prison administrators nationwide to omit federal and state lawbooks from prison library collections. See American Prison Association, Objectives and Standards for Libraries in Adult Prisons and Reformatories, in Library Manual for Correctional Institutions 101, 106-107 (1950). The rise of the prison law library and other legal assistance programs is a recent phenomenon, and one generated largely by the federal courts. See Coyle, supra, at 54-55; B. Vogel, Down for the Count: A Prison Library Handbook 87-89 (1995). See also Ihrig, Providing Legal Access, in Libraries Inside: A Practical Guide for Prison Librarians 195 (R. Rubin D. Suvak eds. 1995) (establishment of law libraries and legal service programs due to "inmate victories in the courts within the last two decades"). Thus, far from recognizing a long tradition of state-sponsored legal assistance for prisoners, Bounds was in fact a major "disruption to traditional prison operation." Vogel, supra, at 87.

The idea that prisoners have a legal right to the assistance that they were traditionally denied is also of recent vintage. The traditional, pre- Bounds view of the law with regard to the State's obligation to facilitate prisoner lawsuits by providing law libraries and legal assistance was articulated in Hatfield v. Bailleaux, 290 F.2d 632 (CA9), cert. denied, 368 U.S. 862 (1961):

"State authorities have no obligation under the federal Constitution to provide library facilities and an opportunity for their use to enable an inmate to search for legal loopholes in the judgment and sentence under which he is held, or to perform services which only a lawyer is trained to perform. All inmates are presumed to be confined under valid judgments and sentences. If an inmate believes he has a meritorious reason for attacking his, he must be given an opportunity to do so. But he has no due process right to spend his prison time or utilize prison facilities in an effort to discover a ground for overturning a presumptively valid judgment.

"Inmates have the constitutional right to waive counsel and act as their own lawyers, but this does not mean that a non-lawyer must be given the opportunity to acquire a legal education. One question which an inmate must decide in determining if he should represent himself is whether in view of his own competency and general prison regulations he can do so adequately. He must make the decision in the light of the circumstances existing. The state has no duty to alter the circumstances to conform with his decision." 290 F.2d, at 640-641.

Consistent with the traditional view, the lower courts understood the Constitution only to guarantee prisoners a right to be free from state interference in filing papers with the courts:

"[A]ccess to the courts means the opportunity to prepare, serve and file whatever pleadings or other documents are necessary or appropriate in order to commence or prosecute court proceedings affecting one's personal liberty, or to assert and sustain a defense therein, and to send and receive communications to and from judges, courts and lawyers concerning such matters." Id., at 637.

See also Oaks v. Wainwright, 430 F.2d 241, 242 (CA5 1970) (affirming dismissal of prisoner's complaint alleging denial of access to library and legal materials on ground that prisoner had not alleged that "he has in any way been denied access to the courts . . ., that he has ever lost the right to commence, prosecute or appeal in any court, or that he has been substantially delayed in obtaining a judicial determination in any proceeding"). Thus, while courts held that a prisoner is entitled to attack his sentence without state interference, they also consistently held that "[p]rison regulations are not required to provide prisoners with the time, the correspondence privileges, the materials or other facilities they desire for the special purpose of trying to find some way of making attack upon the presumptively valid judgments against them." Lee v. Tahash, 352 F.2d 970, 973 (CA8 1965). "If the purpose was not to hamper inmates in gaining reasonable access to the courts with regard to their respective criminal matters, and if the regulations and practices do not interfere with such reasonable access," the inquiry was at an end. Hatfield, 290 F.2d, at 640. That access could have been facilitated without impairing effective prison administration was considered "immaterial." Ibid.

Quite simply, there is no basis in constitutional text, pre- Bounds precedent, history, or tradition for the conclusion that the constitutional right of access imposes affirmative obligations on the States to finance and support prisoner litigation.

II A

Even when compared to the federal judicial overreaching to which we have now become accustomed, this is truly a remarkable case. The District Court's order vividly demonstrates the danger of continuing to afford federal judges the virtually unbridled equitable power that we have for too long sanctioned. We have here yet another example of a federal judge attempting to "direc[t] or manag[e] the reconstruction of entire institutions and bureaucracies, with little regard for the inherent limitations on [his] authority." Missouri v. Jenkins, 515 U.S. at, 126 (THOMAS, J., concurring). And we will continue to see cases like this unless we take more serious steps to curtail the use of equitable power by the federal courts.

Principles of federalism and separation of powers impose stringent limitations on the equitable power of federal courts. When these principles are accorded their proper respect, Article III cannot be understood to authorize the Federal Judiciary to take control of core state institutions like prisons, schools, and hospitals, and assume responsibility for making the difficult policy judgments that state officials are both constitutionally entitled and uniquely qualified to make. See id., at 131-133. Broad remedial decrees strip state administrators of their authority to set long-term goals for the institutions they manage and of the flexibility necessary to make reasonable judgments on short notice under difficult circumstances. See Sandin v. Conner, 515 U.S. 472, 482-483 (1995). At the state level, such decrees override the "State's discretionary authority over its own program and budgets and forc[e] state officials to reallocate state resources and funds to the [district court's] plan at the expense of other citizens, other government programs, and other institutions not represented in court." Jenkins, 515 U.S., at 131 (THOMAS, J., concurring). The federal judiciary is ill equipped to make these types of judgments, and the Framers never imagined that federal judges would displace state executive officials and state legislatures in charting state policy.

Though we have sometimes closed our eyes to federal judicial overreaching, as in the context of school desegregation, see id. at 124-125, we have been vigilant in opposing sweeping remedial decrees in the context of prison administration. "It is difficult to imagine an activity in which a State has a stronger interest, or one that is more intricately bound up with state laws, regulations, and procedures, than the administration of its prisons." Preiser v. Rodriguez, 411 U.S. 475, 491-492 (1973). In this area, perhaps more than any other, we have been faithful to the principles of federalism and separation of powers that limit the Federal Judiciary's exercise of its equitable powers in all instances.

Procunier v. Martinez, 416 U.S. 396 (1974), articulated the governing principles:

"Traditionally, federal courts have adopted a broad hands-off attitude toward problems of prison administration. In part this policy is the product of various limitations on the scope of federal review of conditions in state penal institutions. More fundamentally, this attitude springs from complementary perceptions about the nature of the problems and the efficacy of judicial intervention. Prison administrators are responsible for maintaining internal order and discipline, for securing their institutions against unauthorized access or escape, and for rehabilitating, to the extent that human nature and inadequate resources allow, the inmates placed in their custody. The Herculean obstacles to effective discharge of these duties are too apparent to warrant explication. Suffice it to say that the problems of prisons in America are complex and intractable, and, more to the point, they are not readily susceptible of resolution by decree. Most require expertise, comprehensive planning, and the commitment of resources, all of which are peculiarly within the province of the legislative and executive branches of government. For all of those reasons, courts are ill equipped to deal with the increasingly urgent problems of prison administration and reform. Judicial recognition of that fact reflects no more than a healthy sense of realism. Moreover, where state penal institutions are involved, federal courts have a further reason for deference to the appropriate prison authorities." Id., at 404-405 (footnotes omitted).

State prisons should be run by the state officials with the expertise and the primary authority for running such institutions. Absent the most "extraordinary circumstances," Jones v. North Carolina Prisoners' Labor Union, Inc., 433 U.S. 119, 137 (1977) (Burger, C. J., concurring), federal courts should refrain from meddling in such affairs. Prison administrators have a difficult enough job without federal court intervention. An overbroad remedial decree can make an already daunting task virtually impossible.

I realize that judges, "no less than others in our society, have a natural tendency to believe that their individual solutions to often intractable problems are better and more workable than those of the persons who are actually charged with and trained in the running of the particular institution under examination." Bell v. Wolfish, 441 U.S. 520, 562 (1979). But judges occupy a unique and limited role, one that does not allow them to substitute their views for those in the executive and legislative branches of the various States, who have the constitutional authority and institutional expertise to make these uniquely nonjudicial decisions and who are ultimately accountable for these decisions. Though the temptation may be great, we must not succumb. The Constitution is not a license for federal judges to further social policy goals that prison administrators, in their discretion, have declined to advance.

B

The District Court's opinion and order demonstrate little respect for the principles of federalism, separation of powers, and judicial restraint that have traditionally governed federal judicial power in this area. In a striking arrogation of power, the District Court sought to micromanage every aspect of Arizona's "court access program" in all institutions statewide, dictating standard operating procedures and subjecting the state system to ongoing federal supervision. A sweeping remedial order of this nature would be inappropriate in any case. That the violation sought to be remedied was so minimal, to the extent there was any violation at all, makes this case all the more alarming.

The District Court cited only one instance of a prison inmate having a case dismissed due to the State's alleged failure to provide sufficient assistance, and one instance of another inmate who was unable to file an action. See 834 F. Supp. 1553, 1558, and nn. 37-38 (Ariz. 1992). All of the other alleged "violations" found by the District Court related not to court access, but to library facilities and legal assistance. Many of the found violations were trivial, such as a missing pocket part to a small number of volumes in just a few institutions. Id., at 1562. And though every facility in the Arizona system already contained law libraries that greatly exceeded prisoner needs, the District Court found the State to be in violation because some of its prison libraries lacked Pacific Second Reporters. Ibid. The District Court also struck down regulations that clearly pass muster under Turner v. Safley, 482 U.S. 78 (1987), such as restrictions at some facilities on "brows[ing] the shelves," 834 F. Supp., at 1555, the physical exclusion from the library of "lockdown" inmates, who are the most dangerous and disobedient prisoners in the prison population, id., at 1556, and the allowance of phone calls only for "legitimate pressing legal issues," id., at 1564.

To remedy these and similar "violations," the District Court imposed a sweeping, indiscriminate, and systemwide decree. The microscopically detailed order leaves no stone unturned. It covers everything from training in legal research to the ratio of typewriters to prisoners in each facility. It dictates the hours of operation for all prison libraries statewide, without regard to inmate use, staffing, or cost. It guarantees each prisoner a minimum two-hour visit to the library per trip, and allows the prisoner, not prison officials, to determine which reading room he will use. The order tells ADOC the types of forms it must use to take and respond to prisoner requests for materials. It requires all librarians to have an advanced degree in library science, law, or paralegal studies. If the State wishes to remove a prisoner from the law library for disciplinary reasons, the order requires that the prisoner be provided written notice of the reasons and factual basis for the decision within 48 hours of removal. The order goes so far as to dictate permissible noise levels in law library reading rooms and requires the State to "take all necessary steps, and correct any structural or acoustical problems." App. to Pet. for Cert. 68a.

The order also creates a "legal assistance program," imposing rules for the selection and retention of prisoner legal assistants. Id., at 69a. It requires the State to provide all inmates with a 30-40 hour videotaped legal research course, covering everything from habeas corpus and claims under 42 U.S.C. § 1983 to torts, immigration, and family law. Prisoner legal assistants are required to have an additional 20 hours of live instruction. Prisoners are also entitled to a minimum of three 20-minute phone calls each week to an attorney or legal organization, without regard to the purpose for the call; the order expressly requires Arizona to install extra phones to accommodate the increased use. Of course, legal supplies are covered under the order, which even provides for "ko-rec-type" to correct typographical errors. A Special Master retains ongoing supervisory power to ensure that the order is followed.

The District Court even usurped authority over the prison administrator's core responsibility: institutional security and discipline. See Bell v. Wolfish, 441 U.S., at 546 ("Maintaining institutional security and preserving internal order and discipline" are the central goals of prison administration). Apparently undeterred by this Court's repeated admonitions that security concerns are to be handled by prison administrators, see, e.g., ibid., the District Court decreed that "ADOC prisoners in all . . . custody levels shall be provided regular and comparable visits to the law library." App. to Pet. for Cert. 61a (emphasis added). Only if prison administrators can "documen[t]" an individual prisoner's "inability to use the law library without creating a threat to safety or security" may a potentially dangerous prisoner be kept out of the library, ibid., and even then the decision must be reported to the Special Master. And since, in the District Court's view, "[a] prisoner cannot adequately use the law library under restraint, including handcuffs and shackles," id., at 67a, the State is apparently powerless to take steps to ensure that inmates known to be violent do not injure other inmates or prison guards while in the law library "researching" their claims. This "one free bite" approach conflicts both with our caselaw, see Hewitt v. Helms, 459 U.S. 460, 474 (1983), and with basic common sense. The District Court apparently misunderstood that a prison is neither a law firm nor a legal aid bureau. Prisons are inherently dangerous institutions, and decisions concerning safety, order, and discipline must be, and always have been, left to the sound discretion of prison administrators.

Like the remedial decree in Jenkins, the District Court's order suffers from flaws characteristic of overly broad remedial decrees. First, "the District Court retained jurisdiction over the implementation and modification of the remedial decree, instead of terminating its involvement after issuing its remedy." 515 U.S., at 134 (THOMAS, J., concurring). Arizona correctional officials must continually report to a Special Master on matters of internal prison administration, and the District Court retained discretion to change the rules of the game if, at some unspecified point in the future, it feels that Arizona has not done enough to facilitate court access. Thus, the District Court has "inject[ed] the judiciary into the day-to-day management of institutions and local policies a function that lies outside of our Article III competence." Id., at 135. The District Court also "failed to target its equitable remedies in this case specifically to cure the harm suffered by the victims" of unconstitutional conduct. Id., at 136. We reaffirmed in Jenkins that "the nature of the [equitable] remedy is to be determined by the nature and scope of the constitutional violation." Id., at 88 (majority opinion) (citation and internal quotation marks omitted). Yet, in this case, when the District Court found the law library at a handful of institutions to be deficient, it subjected the entire system to the requirements of the decree and to ongoing federal supervision. And once it found that lockdown inmates experienced delays in receiving law books in some institutions, the District Court required all facilities statewide to provide physical access to all inmates, regardless of custody level. And again, when it found that some prisoners in some facilities were untrained in legal research, the District Court required the State to provide all inmates in all institutions with a 30-40 hour videotaped course in legal research. The remedy far exceeded the scope of any violation, and the District Court far exceeded the scope of its authority.

The District Court's order cannot stand under any circumstances. It is a stark example of what a district court should not do when it finds that a state institution has violated the Constitution. Systemwide relief is never appropriate in the absence of a systemwide violation, and even then should be no broader and last no longer than necessary to remedy the discrete constitutional violation.