PER CURIAM.

In challenging their murder convictions defendants present numerous claims of error. The one that presently engages the Court's attention focuses on the interpretation and use of a polygraph examination report concerning a prosecution witness. In order that we may have a more complete record presented for our considered deliberation of this issue, we remand the cause to the trial court for further proceedings, to be conducted on an expedited basis. We retain jurisdiction to the end that our final decision may be made upon receipt of the trial court's determination on the questions discussed below.

I

The protracted history of this case need be recited only briefly. Defendants, Rubin Carter and John Artis, were charged with homicide of three people in the Lafayette Bar and Grill in Paterson on June 17, 1966. They were convicted in May 1967 on three counts of first degree murder and sentenced to life terms. The judgments of conviction were affirmed by this Court, State v. Carter, 54 N.J. 436 (1969), and the United States Supreme Court denied certiorari, Carter v. New Jersey, 397 U.S. 948, 90 S.Ct. 969, 25 L.Ed.2d 130 (1969).

In 1974 defendants sought a new trial based on the recantation of trial testimony by the State's two key identification witnesses, Alfred Bello and Arthur Bradley. In addition defendants argued that the prosecution had withheld evidence of promises of leniency made to Bello. At the hearing on defendant's motion (recantation hearing) both witnesses repudiated their identification testimony given at trial.

The trial court, after conducting an evidentiary hearing, denied both the motion for a new trial, State v. Carter, 136 N.J. Super. 271 (Cty.Ct. 1974), and subsequent motions for reconsideration, State v. Carter, 136 N.J. Super. 596 (Cty.Ct. 1975). After certifying the case directly this Court vacated the judgments of conviction and ordered a new trial. State v. Carter, 69 N.J. 420 (1976). Although we refused to interfere with the trial court's findings that the recantation testimony was "patently untrue" and "unbelievable," id. at 428, we nevertheless concluded that the State had improperly withheld from defendants exculpatory evidence consisting of proof that promises of protection and favorable treatment had been made to both Bello and Bradley in 1966. Id. at 430-33.

Several months before the second trial in November and December 1976, Bello returned to his 1967 trial testimony, in effect recanting his recantation — of the circumstances surrounding which, particularly the role of polygraphic testing, more below. At the retrial Bello testified consistently with the version he had given at the first trial and again furnished critical identification evidence. Neither the State nor the defense called Bradley as a witness. The jury again convicted defendants of first degree murder and judgments were entered in February 1977. The court sentenced Carter to two consecutive life sentences and one concurrent life sentence. Artis was sentenced to life, with two concurrent life sentences. The Appellate Division, in an unreported opinion, affirmed the judgments of conviction. We granted certification, 84 N.J. 384 (1980).

II

In the second trial, as in the first, Bello's identification testimony constituted a prominent part of the State's case. The version he gave in both proceedings came to be known as the "on-the-street" story. According to this account Bello and Bradley were in the vicinity of the Lafayette Bar and Grill, Paterson, at about 2:30 A.M. on June 17, 1966 bent on effecting a breaking and entry at the Ace Sheet Metal Company, located at the other end of the block from the tavern. While Bradley was attempting to gain entrance into the building, Bello maintained a lookout. As related in our earlier opinions in the case,

Bello ran out of cigarettes and was walking to the tavern to buy some when he heard the shots. As he neared the tavern, two men, Negroes, came toward him, one with a shotgun and the other with a pistol. They were talking loudly and laughing. When Bello realized they were not detectives, he took off and ducked into an alleyway. Presently a white car passed by. Bello recognized it as a 1966 Dodge and noted it had New York license plates. Bello entered the tavern, and seeing the dead and the dying, he went to the cash register to obtain a dime to call the police. The sight of money was too much for Bello who, explaining he was a thief, admitted he scooped some bills from the register. He left the tavern and handed the money to Bradley who, having heard the shots, had started toward the scene. Bello returned to the tavern. He called, or had already called, the police, and when they arrived, he flagged them down.

* * * * * * * *

Both Carter and Artis were questioned early that morning. Both denied involvement. A detective was permitted to testify to each defendant's account of where he had been that night. * * * Both defendants were released that morning. On June 29, twelve days later, both Artis and Carter testified voluntarily before the Grand Jury after signing appropriate waivers.

Defendants were not arrested again until October, when Bello [,] * * * involved in other criminal charges, provided evidence directly incriminating the defendants. Bello said it was Carter who carried the shotgun and Artis who held the pistol; that he recognized both at once when he was confronted by them outside the tavern, and that he had withheld this information because of his own criminal activities that night and his fear of retaliation. [ 69 N.J. at 425-26 (quoting 54 N.J. at 440-41).]

Between 1974 and 1976 Bello offered at least three new versions of this account of the critical events of June 17, 1966. His first recorded deviation came in 1974, some seven years after the first trial, when he gave a statement to an investigator for the Office of the Public Defender. That statement formed the basis for the recantation proceedings, at which Bello repudiated his identification of defendants made at the trial. He testified that although he was indeed "present outside the Lafayette Grill when the shooting took place" and "saw two black men flee the scene," he did not see their faces and "could not identify them." State v. Carter, supra, 69 N.J. at 426-27. He admitted having "lied at the trial when he identified the defendants as the two men." Id. at 427.

Thereafter, two additional variants of Bello's previous renditions surfaced, so at odds with his earlier "on-the-street" versions as to attract the distinguishing label of the "in-the-bar" story. These versions came to light in two statements given in the latter part of 1975 to Assemblyman Hawkins, who was investigating the case on behalf of Governor Byrne in connection with an application for pardon, and in testimony before an Essex County Grand Jury in December 1975. A common ingredient of the "in-the-bar" narrative was that Bello was inside the tavern when two black men — not Carter and Artis — entered through the side door and began shooting; Bello was able to get out of the bar by being "shielded" by a woman who was shot; and as he ran around the corner, he saw Carter and Artis on the sidewalk. In his first statement to Hawkins, Bello insisted that Carter and Artis were unarmed when he saw them on the street. In a second statement to Hawkins and in testimony before the Essex County Grand Jury he modified this account to explain that although defendants were not the triggermen, they were in fact armed. In June 1976 Bello was interviewed by two prosecutor's detectives and repeated essentially the same set of facts he had conveyed to the Grand Jury.

III

Thus it was that as the second trial approached, the State found itself with its key identification witness — hardly a model of rectitude to begin with — in a condition of disarray. It therefore sought to evaluate the credibility of Bello's testimony by subjecting him to a polygraph examination on August 7, 1976. The test was conducted by Leonard H. Harrelson of Leonarde Keeler, Inc., a polygraph institute in Chicago. In arranging for the examination Prosecutor Humphreys wrote to Professor Harrelson, emphasizing that the State's interest was only in "ascertaining the truth and prosecuting the guilty, not the innocent," and assuring the polygraphist that the prosecution had "absolutely no interest or bias in the outcome of these tests, or this case, other than to see that the truth is revealed, justice is done and the chips fall where they may."

Professor Harrelson submitted a written report dated August 24, 1976. A copy was furnished to defense counsel on September 14, 1976. The report summarizes the examination and concludes:

After careful analysis of this subject's polygrams, it is the opinion of the examiner that his 1966 [ sic: 1967] testimony at trial was true, and that the statement recanting his original statement is not true.

However, this conclusion is shrouded in considerable confusion, generated in no small measure by the facts, later developed, that Harrelson had never familiarized himself with Bello's 1967 trial testimony and no member of the prosecution team had informed him that at the first trial Bello had related the "on-the-street" version. From information supplied after trial it seems likely that what Harrelson intended to communicate was that Bello's testimony at the recantation hearing indicated he was unable to identify Carter and Artis was false, but that his statements to Harrelson that he was in the bar before and at the time of the shootings and first saw Carter and Artis outside the bar after the shootings were true. Significant in this connection is a January 1978 affidavit of Harrelson, stating categorically that after testing Bello he spoke with Chief Assistant Prosecutor Kayne "and related to him, among other things, that Alfred Bello told me during the examination that he was in the Lafayette Bar prior to and at the time of the shootings. I told [Kayne] that in my opinion Bello was telling me the truth."

The potential for misunderstanding arising from all of this is compounded by Mr. Kayne's admission, in a June 1978 affidavit, that "[a]fter the tests had been completed, Professor Harrelson mentioned to [Acting Chief of Detectives] DeSimone (and perhaps to Prosecutor Humphreys by phone) a preliminary view that Bello was in the bar at the time of the shooting." Furthermore, Kayne says that "[a]t the conclusion of the examinations, Professor Harrelson advised us that because of Bello's heart irregularities, accurate interpretation of Bello's chart was difficult; and he would, therefore, have to review his charts and furnish us his final report later." However, nothing further appears in the record bearing on the possible "preliminary" nature of Harrelson's oral report. His written report does not speak to the point. His post-trial affidavit may reasonably be taken to suggest that if any review was conducted, the conclusion nevertheless remained the same, namely, that Bello's "in-the-bar" version was true. Despite this, Harrelson's written report, which the State did furnish to defense counsel, contains the now-challenged and seemingly repudiated conclusion that Bello's 1967 trial testimony — the "on-the-street" account — was true.

IV

Defendants contend that the apparent error in Harrelson's conclusion, undisclosed to them until after trial, assumes critical significance when considered in light of the use to which the test results may have been put. After receipt of Harrelson's report two representatives of the prosecution interviewed Bello on September 14 and 15, 1976; and while the record is far from clear on the use, if any, made of the polygraph report during that confrontation, the fact is that at the conclusion thereof Bello retreated from his "in-the-bar" story and returned to his 1967 account. As we have indicated, Bello's testimony at the second trial was in keeping with this "on-the-street" account.

The thrust of defendants' argument is that even though the State did not introduce the polygraph examination and report into evidence, the circumstances surrounding the polygraph procedure and the use made of the report were of "controlling importance" at the trial. Defense counsel contend that had they known at the time of trial that which they came to know only after the trial, that is, that the lie detector report meant the opposite of what it clearly said, they would have turned the evidence to the defendants' advantage. They argue that they could have impeached Bello's testimony and, perhaps more significantly, shown that Bello had changed his story through prosecutorial deceit and coercion. Defendants' attorneys point out that they could also have cross-examined DeSimone on the manner in which he brought Bello back to the "on-the-street" version. They suggest that Bello was particularly vulnerable to threats of prosecution for perjury for his "in-the-bar" account given in 1975 to an Essex County Grand Jury. They insist, moreover, that access to these trial stategems was denied them by a "prosecution-induced" trial court ruling that questioning directed to the State's use of the Harrelson test to "turn Bello around" would open the door to admission of the purported test results and to Harrelson's testimony supposedly supporting the truth of Bello's "on-the-street" story. Hence the defense complains they were "hamstrung" by the trial court's warning that questions probing the reasons for Bello's change in testimony might permit a discussion of the lie detector results which, on their face, were injurious to defendants but which, in fact, might have been helpful to them had they been told the truth by the prosecution.

V

Our concern at this juncture is whether a new trial is warranted on either of two grounds. The first inquiry is whether the prosecution withheld material evidence favorable to the defendant, thereby violating the rule developed in Brady v. Maryland, 373 U.S. 83, 83 S.Ct. 1194, 10 L.Ed.2d 215 (1963), and its progeny. See United States v. Agurs, 427 U.S. 97, 96 S.Ct. 2392, 49 L.Ed.2d 342 (1976); Moore v. Illinois, 408 U.S. 786, 92 S.Ct. 2562, 33 L.Ed.2d 706 (1972); Giglio v. United States, 405 U.S. 150, 92 S.Ct. 763, 31 L.Ed.2d 104 (1972). If a Brady violation is not established, there nevertheless remains the question whether the evidence accumulated since the trial, taken together with the facts to be developed at the hearing on remand, satisfies the test for a retrial based on newly discovered evidence.

Relying on the Supreme Court ruling in Brady, supra, defendants charge the prosecution with having withheld material evidence favorable to them in connection with the Harrelson polygraph. They contend that the withholding of such evidence is a denial of due process and the right to a fair trial, irrespective of the good faith or bad faith of the prosecution. See Brady, supra, 373 U.S. at 87, 83 S.Ct. at 1196, 10 L.Ed.2d at 218. "When the `reliability of a given witness may well be determinative of guilt or innocence,' nondisclosure of evidence affecting credibility falls within [the Brady] rule." Giglio, supra, 405 U.S. at 154, 92 S.Ct. at 766, 31 L.Ed.2d at 108 (quoting Napue v. Illinois, 360 U.S. 264, 79 S.Ct. 1173, 3 L.Ed.2d 1217 , 1221 (1959)). The defense relies as well on this Court's emphasis in State v. Carter, supra, 69 N.J. at 433, on the State's obligation to disclose to defense counsel information it possesses or materials in its file — an obligation that is "not limited to evidence that affirmatively tends to establish a defendant's innocence but would include any information material and favorable to a defendant's cause even where the evidence concerns only the credibility of a State's witness." Id.

In order to establish a Brady violation the defense must demonstrate that (1) the prosecution failed to disclose the evidence; (2) the evidence was of a favorable character for the defense; and (3) the evidence was material. See Moore v. Illinois, supra, 408 U.S. at 794-95, 92 S.Ct. at 2567-68, 33 L.Ed.2d at 714. Because of the many kinds of evidence that can be considered favorable to a defendant, the standard of materiality for determining whether a Brady violation is established varies according to which of three general factual settings exists. First, where the prosecution has knowingly used perjured testimony, the conviction will be set aside if there is a reasonable likelihood that the false testimony could have affected the judgment of the jury. United States v. Agurs, supra, 427 U.S. at 103, 96 S.Ct. at 2397, 49 L.Ed.2d at 349-50. The second situation, illustrated by Brady itself, is one in which the defense has made a specific request for the suppressed evidence. If that evidence might have affected the outcome of the trial, then the judgment must be vacated. Id. at 104, 96 S.Ct. at 2397, 49 L.Ed.2d at 350. The third variation is found where the defense has made no request, or only a general one, for the undisclosed material, in which case the defense must show that "the omitted evidence creates a reasonable doubt that did not otherwise exist." Id. at 112, 96 S.Ct. at 2401, 49 L.Ed.2d at 355. This sliding scale analysis has received wide acceptance. See Talamante v. Romero, 620 F.2d 784, 787-88 (10th Cir. 1980); United States v. Imbruglia, 617 F.2d 1, 4-5 (1st Cir. 1980); United States v. Provenzano, 615 F.2d 37, 47 (2d Cir. 1980); Monroe v. Blackburn, 607 F.2d 148, 150-51 (5th Cir. 1979); United States v. Librach, 602 F.2d 165, 167 (8th Cir. 1979); Galtieri v. Wainwright, 582 F.2d 348, 363 n. 27 (5th Cir. 1978); United States v. Jackson, 579 F.2d 553, 559-60 (10th Cir. 1978); Ostrer v. United States, 577 F.2d 782, 786 (2d Cir. 1978); United States v. McCrane, 575 F.2d 58, 61 (3rd Cir. 1978); United States v. Anderson, 574 F.2d 1347, 1353 (5th Cir. 1978); United States ex rel. Marzeno v. Gengler, 574 F.2d 730, 735-36 (3rd Cir. 1978); United States v. Weidman, 572 F.2d 1199, 1203-04 (7th Cir. 1978); United States v. Dansker, 565 F.2d 1262, 1264 (3rd Cir. 1978); United States v. Brown, 562 F.2d 1144, 1149 (9th Cir. 1977); Cannon v. Alabama, 558 F.2d 1211, 1213 n. 2 (5th Cir. 1977); Wagster v. Overberg, 560 F.2d 735, 739 (6th Cir. 1977); United States v. McCrane, 547 F.2d 204, 205 (3rd Cir. 1976); Garrison v. Maggio, 540 F.2d 1271, 1273 (5th Cir. 1976). See also Note, 94 Harv. L.Rev. 887, 892-93 (1981).

In the present case it appears undisputed that the defense made a specific request for all information concerning Bello's polygraph tests. Therefore, if it is determined that the prosecution did in fact suppress information favorable to the defense, the question of materiality for Brady purposes becomes whether that evidence "might have affected the outcome of the trial." United States v. Agurs, supra, 427 U.S. at 104, 96 S.Ct. at 2397, 49 L.Ed.2d at 350 . In general, a conviction need not be vacated under the Brady rule if the undisclosed information is merely cumulative or repetitious as to the purpose for which it could have been used. See DeMartino v. Weidenburner, 616 F.2d 708, 711 (3rd Cir. 1980).

Additionally, for the Brady rule to apply there must be a threshold determination of prosecutorial knowledge. To meet this requirement the defense must demonstrate that the evidence it charges the prosecution with having suppressed "had been known to the prosecution but unknown to the defense." United States v. Agurs, supra, 427 U.S. at 103, 96 S.Ct. at 2397, 49 L.Ed.2d at 349. The Supreme Court has indicated that although for Brady purposes such prosecutorial knowledge need not be actual, the contested evidence must at least be "in [the prosecutor's] file." Id. at 110, 96 S.Ct. at 2400, 49 L.Ed.2d at 353.

If it is determined that there is no Brady violation, then the focus shifts to a newly discovered evidence analysis — that is, the information acquired by the defense since trial would have to be assessed in light of the requirements for a new trial based on newly discovered evidence. The inquiry then becomes whether the post-trial revelation that Harrelson's report did not mean what it purported to say constitutes such newly discovered evidence as to justify a new trial.

Whereas the test of materiality for the granting of a new trial under a Brady analysis is simply whether the suppressed evidence might have affected the outcome of the trial, ante at ____, the test to be satisfied under a newly discovered evidence approach is more stringent. This Court has stated repeatedly that to qualify as newly discovered evidence entitling a party to a new trial, the new evidence must be (1) material to the issue and not merely cumulative or impeaching or contradictory; (2) discovered since the trial and not discoverable by reasonable diligence beforehand; and (3) of the sort that would probably change the jury's verdict if a new trial were granted. State v. Artis, 36 N.J. 538 , 541 (1962); see State v. Puchalski, 45 N.J. 97, 107 (1965); State v. Sullivan, 43 N.J. 209 , 233 (1964); State v. Johnson, 34 N.J. 212 , 222 (1961); State v. Smith, 29 N.J. 561 , 573 (1959); State v. Richter, 21 N.J. 421, 423 (1956); State v. Haines, 20 N.J. 438 , 444 (1956); State v. Vaszorich, 13 N.J. 99 , 130 (1952); State v. Bunk, 4 N.J. 482, 486 (1950). All three tests must be met before the evidence can be said to justify a new trial. State v. Johnson, supra, 34 N.J. at 223. Cf. Quick Chek Food Stores v. Springfield Tp., 83 N.J. 438 , 445 (1980) (motion for new trial in civil case should be granted "when that evidence would probably alter the judgment and by due diligence could not have been discovered before the court announced its decision."). See also Pressler, Current N.J. Rules, Comment R. 3:20-1.

In the record now before us the arguments directed to these important issues are based largely on ex parte affidavits, together with reports made before and after trial. Such a record does not lend itself to suitable appellate resolution. A more precise understanding of the pertinent facts is obtainable only through live testimony and a trial level determination in the first instance.

On the remand the trial court is directed to take further testimony and make findings of fact regarding the prosecution's knowledge of the ambiguity contained in Harrelson's report. It is clear that after examining Bello, Harrelson told prosecution representatives that he believed Bello was telling the truth when he said he was in the tavern before and at the time of the shootings. Whether this was so tentative an expression of opinion as to justify the prosecution's failure to disclose it cannot be divined. In any event Harrelson's correspondence does not explicitly verify the prosecution's contention that this was only a "preliminary" report. Does Harrelson concede that his oral report was preliminary? Did he indicate that upon further review his initial conclusion might be altered? Did he suggest to the prosecution team members in any fashion that the conclusions in his written report were at variance with what he had told them in his oral report, namely, that Bello's "in-the-bar" version was the truth?

An additional area ripe for exploration at the evidentiary hearing on remand is precisely what use the prosecution made of Harrelson's report in confronting Bello on September 14 and 15, 1976, prior to the second trial. The State argues that at the September 1976 interviews, Bello gave Chief Assistant Prosecutor Kayne and Chief DeSimone a detailed account of the various pressures and inducements that resulted in his recantation, explained the basis for his statements to Assemblyman Hawkins and the Essex County Grand Jury, and thereupon returned to his 1967 trial version of the shootings. During the course of the 1976 trial, both Prosecutor Humphreys and Assistant Prosecutor Marmo told the court that the lie detector tests were instrumental in getting Bello back to the "on-the-street" version. The hearing on remand should explore how the polygraphist's report was linked to that result.

We anticipate that the following people, or at least some of them, will testify at the evidentiary hearing: Messrs. Bello, Harrelson, Humphreys, Kayne, Marmo and defense counsel. Through the process of examination and cross-examination the parties can create a record supportive of their respective positions on the issues outlined herein. On the completion of that testimony the trial court should make findings of fact bearing on those issues. First, it should assess those facts, together with all other relevant and material evidence disclosed since trial, in the light of the requirements for establishing a Brady violation as underscored herein, ante at 312. Of course if the court finds no prosecutorial knowledge, then there cannot be a Brady violation, irrespective of the favorable character of the evidence or its materiality in the Brady sense. Nonetheless, we direct that even if the requirement of prosecutorial knowledge is determined not to be satisfied, the trial court should complete its findings and analysis under Brady for the sake of completeness of the record on review and final disposition in the event of a successful defense challenge to any threshold finding of no prosecutorial knowledge. For the same reasons — that is, creation of a full and complete record appropriate for final appellate determination — the trial court should continue beyond the point of deciding whether there was a Brady violation and assess the impact of the facts on the newly discovered evidence test as set forth herein.

The legal conclusion reached by this process of analysis — new trial or no new trial under either a Brady or newly discovered evidence approach — should be stated with supporting reasons. The entire supplemental record should thereafter be filed promptly with the Clerk of the Supreme Court for our review.

As a final word we emphasize that in remanding the cause for the limited purpose set forth in this opinion, we intimate no view on the ultimate merits of the issues discussed herein or of any of defendants' numerous other claims of error.

Remanded. Jurisdiction remains with this Court. For remandment — Chief Justice WILENTZ and Justices SULLIVAN, PASHMAN, CLIFFORD, SCHREIBER, HANDLER and POLLOCK — 7.

Opposed — none.