OPINION OF THE COURT
Ronald A. Zweibel, J.Defendant Melvin Graham moves pursuant to CPL 440.10 *418(1) (h) for an order vacating his judgment of conviction of robbery and felony murder on grounds that his US Constitution Sixth Amendment right to confrontation was violated by the admission of his nontestifying codefendant’s confession at their joint trial.
Defendant raised this issue previously on direct appeal. On May 12, 1986, the Appellate Division, Second Department, affirmed defendant’s judgment of conviction, holding that defendant’s own confession "interlocked” with his codefendant’s, and therefore it was "unlikely that any prejudice to this defendant resulted from the admission of his codefendant’s confession” (People v Graham, 120 AD2d 611). Leave to appeal to the Court of Appeals was subsequently denied. (68 NY2d 812.)
Under CPL 440.10 (2) (a), a court must deny a motion to vacate judgment when "[t]he ground or issue raised upon the motion was previously determined on the merits upon an appeal from the judgment”.
There is an exception to this rule for cases where there has been an intervening "retroactively effective change in the law controlling such issue” (CPL 440.10 [2] [a]).
Defendant argues that Cruz v New York (481 US 186) created just such a retroactively effective change in the law since his appeal became final.
BACKGROUND OF CRUZ v NEW YORK
In Bruton v United States (391 US 123), the Supreme Court held that a defendant is deprived of his Sixth Amendment right to confrontation when a nontestifying codefendant’s confession implicating the defendant is admitted at their joint trial. Overruling prior case law, the court found that a limiting instruction is inadequate protection because of the likelihood that jurors would be unable to obey such an instruction and that prejudice would thereby result.
Shortly after Bruton (supra), the New York Court of Appeals carved out an exception to the Bruton rule for "interlocking confessions”. The court found that " 'where the jury [hears] not only a codefendant’s confession but the defendant’s own confession no such "devastating” risk attends the lack of confrontation as was thought to be involved in Bruton.’ ” (People v McNeil, 24 NY2d 550, 553.)
The Supreme Court did not confront the "interlocking confession” issue until some 10 years after McNeil (supra) in *419Parker v Randolph (442 US 62). Unfortunately, in a 4-to-4 decision, the court was unable to resolve the issue.
After Parker (supra), the New York Court of Appeals continued to follow McNeil (supra). (See, e.g., People v Smalls, 55 NY2d 407, 415; People v Berzups, 49 NY2d 417, 425.)
The Supreme Court resolved the question in Cruz v New York (481 US 186, supra) by holding that the admission of interlocking confessions may violate a defendant’s confrontation rights because once jurors heard the codefendant’s confession, they would be unable to obey the instruction not to use the codefendant’s confession as evidence against the defendant. The court noted (481 US, at 193) that this conclusion did "no more than reaffirm” the central proposition of Bruton (supra). The court further held that the defendant’s own confession may be considered in determining whether the violation was harmless. (Cruz v New York, supra, 481 US, at 193-194.)
mSTORY OF RETROACTIVITY
In order to determine the retroactivity of Cruz v New York (supra) it is necessary to understand the historical underpinnings and changes in the law of retroactivity. In Linkletter v Walker (381 US 618) the Supreme Court was faced with the issue of deciding whether the newly created exclusionary rule applied to cases already final before the Mapp decision (Mapp v Ohio, 367 US 643). Noting that the Constitution did not mandate either prospective or retrospective application of newly announced rules, the court observed that changes in the law were historically given retroactive effect in cases on direct review. In contrast, the determination of whether to apply a rule retroactively to cases already final depended on the characteristics of the particular rule in question. (Linkletter v Walker, supra, at 627.)* The court provided three factors to be considered in making this determination on cases already final: the history of the rule in question, the purpose to be served by the new standard, and whether retroactive application would further or retard its operation. With regard to these factors, the court further noted that where the purpose of a newly announced standard "went to the fairness of the trial — the very integrity of the fact-finding process” — the deci*420sion announcing the new standard would routinely be applied fully retroactively even to cases already final. (Linkletter v Walker, supra, at 639.)
The court then proceeded to apply the balancing test to the case before it. Finding that the purpose of the exclusionary rule was to deter improper police conduct and bore little relation to the integrity of the trial, the court denied retroactive application of the Mapp decision to cases already final.
The next year, in Tehan v Shott (382 US 406), the court again refused to apply a newly created rule to a case already final. This case involved the US Constitution Fifth Amendment right prohibiting the government from commenting on defendant’s failure to testify. Using the Linkletter factors, the court denied retroactive application of the new rule to cases already final.
The same year as Tehan (supra), in Johnson v New Jersey (384 US 719), the court disregarded what it had said the year before in Linkletter (supra) about new rules being traditionally applied retroactively to cases still on direct review, and used the balancing test to deny retroactive application of a new decision not only to cases already final, but also to cases not yet final. The court also modified its previous position on rules affecting the fairness of the trial or the integrity of the fact-finding process. The court examined the "degree” to which the newly created rules relating to voluntariness of confession affected the fact-finding process. Finding that while the new rule did affect the fairness of the trial to some degree it did not do so to such a degree as to outweigh the considerations of reliance and effect on administration of justice, the court denied retroactive application. (Johnson v New Jersey, supra, at 728-729.)
The next year, in Stovall v Denno (388 US 293), the court applied the balancing test to a rule involving the right to counsel. Applying a slightly modified version of the Linkletter factors, the court balanced "the purpose to be served by the new standards * * * the extent of the reliance * * * on the old standards, and * * * the effect on the administration of justice of a retroactive application”. (388 US, at 297.) The court concluded that even though the rules in question the Wade-Gilbert right to counsel did relate to the integrity and reliability of the identification process, the "unusually strong” reliance and burden factors mandated that the new rules be applied prospectively only. The court refused to distinguish between cases already final and those still on direct review.
*421In contrast, one year later the court exhibited little difficulty in determining that the rule announced in Bruton v United States (391 US 123, supra) ought to be given retroactive application to cases already final because a violation of a defendant’s right to confrontation goes " 'to the basis of fair hearing and trial’ ” (Roberts v Russell, 392 US 293, 294 [1968]). Applying the Stovall balancing test, the court concluded that "Due regard for countervailing considerations— reliance on the old standard * * * and the impact of retroactivity upon the administration of justice * * * does not counsel against retroactivity of Bruton * * * And even if the impact of retroactivity may be significant, the constitutional error presents a serious risk that the issue of guilt or innocence may not have been reliably determined.” (392 US, at 295.)
Similarly, the next year in another Confrontation Clause case, the court granted retroactivity to a case already final because it found that inability to cross-examine could have a significant impact on the fact-finding process, and would thus outweigh the other considerations. (Berger v California, 393 US 314, granting retroactivity to Barber v Page, 390 US 719.)
A few months later, in Desist v United States (394 US 244), the court declined to apply the exclusionary rule of Katz v United States (389 US 347) retroactively to cases still on direct review. Using the Stovall factors, the court found the purpose of the Katz rule to be deterrence of improper police conduct; since the misconduct had already occurred the purpose of the rule would not be enhanced by retroactive application. In addition, the court for the first time referred to the fact that Katz, which explicitly overruled a prior Supreme Court precedent, was a "clear break” with past decisions, as an additional factor weighing against retroactivity.
In a lengthy dissent, Justice Harlan called for a full reevaluation of retroactivity analysis. Admitting that he himself had been one of the Justices who had supported case-by-case retroactivity analysis as a means of dealing with the outpouring of new decisions in the 1960’s, he now took the position that correct principles of judicial review required that newly announced rules be applied to all cases which came up on direct review. Justice Harlan found no authority to permit a court to apply a new rule only to the defendant in the case in which it was announced, and then decide other cases on direct appeal according to the disapproved old standard. In addition, *422giving the benefit of the new rule only to the defendant whose case had fortuitously been selected as the vehicle for announcing the new standard created indefensible inequities in the treatment of identically situated defendants.
With regard to convictions already final coming up on collateral review, however, Justice Harlan took the position, based largely on his view of the scope of habeas corpus, that courts were simply required to ensure that such defendants had been given the benefit of the law as it stood at the time of their convictions. Newly announced decisions should not be applied retroactively to cases already final.
Justice Harlan noted that before denying retroactive application of a rule to a case already final, it would first be necessary to determine whether the decision in question actually did create a "new” rule, or whether it "simply applied a well-established constitutional principle to govern a case which is closely analogous to those which have been previously considered in the prior case law.” (Desist v United States, 394 US, at 263, supra [Harlan, J., dissenting opn].) If the decision was determined to be merely an application of an old standard to a new fact pattern, then it would be applied retroactively to cases already final, since no new rule was actually created.
Two years later, in Williams v United States (401 US 646), the court declined to accept Justice Harlan’s invitation, and repeated its position that it saw no reason to differentiate between cases on direct review and collateral attacks of convictions already final. Justice Harlan, dissenting in Williams and concurring in its companion case, Mackey v United States (401 US 667), reiterated the approach he had taken in Desist (supra). He stressed the need for finality in the operation of the criminal process as reason for denying retroactive application of new decisions to cases already final, and explained again the inappropriateness of refusing to apply decisions retroactively to cases arising on direct appeal.
Justices Marshall, also writing separately in these cases, agreed with Justice Harlan that cases on direct review should always receive the benefit of a newly declared rule. He disagreed with Justice Harlan, however, on the proper standard to use for cases already final. Justice Marshall took the position that the retroactivity determination for cases already final should be made through use of the Stovall balancing test. (Williams v United States, supra, 401 US, at 665.)
*423The next year, in Adams v Illinois (405 US 278), the court used the balancing test to determine whether the new rule granting the right to counsel at preliminary hearings should be applied retroactively to a case not yet final. Finding that this right to counsel differed from the right to counsel at trial in the "degree” of its impact on the integrity of the fact-finding process, the court applied the balancing test rather than automatic retroactivity. The court then denied retroactive application.
Justices Marshall and Douglas both dissented, pointing out the inequity of applying a rule retroactively only to the defendants in one case and denying its benefit to other defendants whose cases were also on direct review. Justice Douglas also found the attempts to distinguish between cases which had a direct impact on the integrity of the fact-finding process and those which did not " 'intractable’ ” (Adams v Illinois, supra, at 287).
The court granted full retroactivity the following year to its holding in In re Winship (397 US 358), that the standard of proof beyond a reasonable doubt was constitutionally mandated (Ivan V. v City of New York, 407 US 203). The court found that the purpose of the standard was to overcome an aspect of the trial that substantially impaired the integrity of the fact-finding process. The court cited its pronouncements in prior cases, including Roberts v Russell (supra), that under s’xch circumstances full retroactivity was required, no matter how weighty the second two Stovall factors.
Three years later, relying on Ivan V. v City of New York (supra), the court held that the new prohibition against shifting the burden of proof on an element of a crime (Mullaney v Wilbur, 421 US 684) would be granted full retroactivity (Hankerson v North Carolina, 432 US 233). The court reiterated its position that neither justifiable reliance by a State on an old rule nor potentially devastating impact of retroactive application on the administration of justice could outweigh a rule which had as its major purpose protecting the integrity of the fact-finding process at trial. The court conceded that in certain previous cases it had used the second and third Stovall factors to deny retroactivity to a new rule which bore some relation to the fact-finding process, but denied that it had ever done so when the major purpose of the new rule was " 'to overcome an aspect of the criminal trial that substantially impairs its truth-finding function’ ” (432 US, at 243.)
*424Concurring in the judgment, Justices Marshall and Powell both called for adoption of the Harlan approach calling for retroactive application to all cases still on direct appeal.
In 1982, the court accepted Justice Harlan’s approach to retroactivity for cases still on direct review, at least for decisions construing US Constitution Fourth Amendment, but did not adopt his view as to rules applicable to cases already final. (United States v Johnson, 457 US 537.)
In reviewing the history of retroactivity analysis since Linkletter (381 US 618, supra), the court in Johnson (supra) pointed out that there were two categories of cases which were routinely given full retroactive effect even to cases already final. The first category included decisions which "merely * * * applied settled precedents to new and different” fact patterns. (United States v Johnson, supra, at 549.) Since such decisions did not in fact alter the preexisting rule in any material way, they were given full retroactive effect. The second category which had the effect of declaring a conviction void, such as decisions applying the principles of double jeopardy. Because such cases affected the court’s jurisdiction, they too were always applied retroactively.
Two years after Johnson (supra), in Solem v Stumes (465 US 638), a case on collateral review, the court declined to follow the second part of Justice Harlan’s approach. The court found that the rules about waiver of right to counsel announced in Edwards v Arizona (451 US 477) would not be applied retroactively to cases already final. This conclusion was arrived at by applying the Stovall balancing test.
A year later, the court was faced with the correlative question of the retroactivity of Edwards (supra) to cases not yet final. The court reaffirmed that it was now following Justice Harlan’s approach for cases still on direct review, and that therefore Edwards would apply retroactively to nonfinal cases (Shea v Louisiana, 470 US 51).
The next year, in Allen v Hardy (478 US 255), the court again denied retroactive application of the rule, requiring prosecutors to give reason for their peremptory challenges (Batson v Kentucky, 476 US 79) to cases already final. The court found that the rule in Batson served many values, only one of which, the ensuring of a neutral jury, was related to the fact-finding process at trial, and thus concluded that the new rule was not one which would have "such a fundamental impact on the integrity of factfinding as to compel retroactive *425application.” (478 US, at 259.) Terming Batson a "clear break” with past decisions, the court found that the factors of reliance on the old standard which Batson overruled and the impact of retroactive application were significant and on balance weighed in favor of nonretroactivity.
When the court was faced with determining whether Batson (supra) should be applied retroactively to cases which were not yet final, the court once more stated its adoption of the first part of Justice Harlan’s approach, and held that Batson would be applied retroactively to cases still on direct review. (Griffith v Kentucky, 479 US 314.) The court went on, for the first time, to set out a "bright-line rule” for retroactivity: all new rules would be applied retroactively to all cases not yet final, with no exception even for rules which were a "clear break” from past decisions. While rejecting the balancing test for cases on direct review, the court commented in passing that the Stovall factors might be useful in determining whether convictions already final should get the benefit of a "new rule”.
In the court’s most recent retroactivity decision, Yates v Aiken (484 US 211, 108 S Ct 534), the court did not reach the government’s suggestion that the court adopt the Harlan approach to retroactivity for cases already final. The court explained that it had no need to make the decision in the case before it, because the alleged "new rule” was merely an application of the previously announced standard to slightly different facts. Since there is no new rule created, retroactivity analysis is unnecessary; the decision is given full retroactive application.
Thus, at present, retroactivity analysis for cases still on direct review has been greatly simplified, as new rules will now be applied to all cases not yet final. Retroactive application of new rules to cases already final, however, continues to be determined by the Stovall balancing test unless the rule in question falls into one of the exceptions described in Johnson (supra).
RETROACTIVITY OF CRUZ v NEW YORK
In this case we are faced with the question of whether Cruz v New York (481 US 186, supra) should apply to a conviction which is already final. The Supreme Court characterized its decision in Cruz v New York as "no more than [a] reaffirm[ation of the] central proposition” of Bruton. (481 US, *426at 193.) Cruz can thus be seen as merely an application of the Bruton rule to a new fact pattern, i.e., to the situation of interlocking confessions. A decision which merely applies settled precedents to new fact patterns is applied retroactively to cases already final, since the new rule does not actually change the previously existing law. (See, Yates v Aiken, 484 US 211, 108 S Ct 534, supra.) Thus, under this analysis, Cruz applies retroactively to a case like defendant’s which was final when the Cruz decision was announced.
Assuming, arguendo, that the Cruz decision does not fall into this category, the application of traditional retroactivity analysis leads to the same result. The Bruton rule applied in Cruz v New York (supra) was created to ensure that a defendant’s right to confrontation or cross-examination would not be violated at trial. The right of cross-examination, inherent in the right to confrontation, assures " 'the accuracy of the truth-determining process’ ” (see, Chambers v Mississippi, 410 US 284, 295; Dutton v Evans, 400 US 74, 89; Bruton v United States, 391 US 123, 135-137, supra). Denial or diminution of this right calls into question the ultimate " ' "integrity of the fact-finding process” ’ ” (Chambers v Mississippi, supra, at 295, citing Berger v California, supra, 393 US, at 315).
Indeed, Bruton (supra) itself was given full retroactivity. (Roberts v Russell, 392 US, at 294-295, supra.) The court in holding Bruton fully retroactive held that whatever the weight of the second two Stovall factors, they do not overbalance the need for retroactive application of a rule intended to prevent an error which impinges on the truth-finding process at trial. (Compare, Ivan V. v City of New York, 407 US 203, supra, and Hankerson v North Carolina, 432 US 233, supra, with Allen v Hardy, 478 US 255, supra, and Solem v Stumes, 465 US 638, supra.)
Thus, under either analysis, Cruz v New York (supra) merits retroactive application to cases already final. This court therefore agrees with defendant’s claim that there has been a retroactively effective change in the law controlling the issue he previously raised on appeal, and will thus proceed to address the merits of defendant’s motion.
MERITS
THE FACTS OF DEFENDANT’S CASE
Defendant Melvin Graham, Benjamin Stephens and Darryl Green were indicted for murder and robbery as the result of *427the fatal shooting of Ronald Sanders in a subway station. Defendant and Stephens were tried jointly; Green was tried separately.
Neither defendant nor Stephens took the stand at their joint trial. At that trial, the People introduced signed written statements that both defendants had given to the police, as well as a videotaped confession by each man.
In his own statements, defendant explained that he had been in a fast-food shop when Darryl "Rocky” Green came up and pointed out Ronald Sanders, who was wearing gold rings. Green informed defendant that he knew that Sanders had money on him, and said "let’s get this man.” When Sanders left, defendant and Green followed him. By this point, defendant had drunk "a few beers.” They caught up with Sanders on the street, and defendant ordered him to stop. When Sanders refused to stop, defendant pulled out his gun and shot at Sanders three times, but missed.
Sanders ran down the stairs into a subway station. According to defendant, it was at this point that codefendant Stephens came along. The three men then pursued Sanders into the subway station. Defendant encountered Sanders on the subway platform using the telephone. Sanders said "I’m tired of this shit” and charged at defendant. Defendant responded by ordering Sanders to back up; when Sanders refused, defendant shot at him twice. Defendant heard Sanders yell, and then ran up the stairs out of the subway.
Codefendant Stephens’s statement interlocked with defendant’s. Stephens confirmed that it was Green who had pointed out Sanders, Green who had stated that Sanders had money and Green who had announced that he was going to get Sanders. Stephens’s version of the street encounter was similar to that of defendant, although he omitted the fact that defendant had told Sanders to stop before firing at him.
Stephens described defendant as "drunk on beer.” In his statement, Stephens omitted mention of Sanders having charged at defendant before defendant fired at him near the telephone. Stephens claimed that defendant had shot Sanders at close range in the chest, information not contained in defendant’s own statement.
Stephens supported defendant’s assertion that he had run out of the subway immediately after firing the shots. After defendant left, Stephens and Green observed Sanders run over to the token booth, cry for help, and then fall to the ground. *428According to Stephens, Green took a ring from Sanders’s hand, began to leave, and then ran back and went through Sanders’s pockets. The two men then ran out.
Stephens and Green encountered defendant on the street, and told him that the man he had shot was dying. When defendant expressed disbelief that he actually hit Sanders, Stephens told him to go down into the subway to see for himself. Defendant did so and returned to say that the token booth was surrounded by police.
Defendant was identified in court by the token booth clerk.
HARMLESS ERROR
A defendant’s right to confrontation is violated by the admission, at their joint trial, of his nontestifying codefendant’s confession incriminating him, even if the jury is given limiting instructions, and even if his own confession is admitted against him as well. (Cruz v New York, 481 US 186, supra.) Defendant’s right to confrontation was therefore violated in this case, and his conviction must be reversed unless the error is found to be harmless beyond a reasonable doubt. (Harrington v California, 395 US 250 [1969].) For a constitutional error to be harmless, proof of guilt must be overwhelming and there must be no reasonable possibility that the jury would have acquitted the defendant but for the error. (Harrington v California, 395 US 250, supra; People v Crimmins, 36 NY2d 230.) A defendant’s own confession may be considered in assessing whether the violation of the Confrontation Clause was harmless. (Cruz v New York, supra, 481 US, at 193-194.)
In the instant case, the error occasioned by the denial of severance and the violation of the Bruton rule does not require reversal. Defendant’s own statements were presented to the jury in writing and on videotape, as were his codefendant’s. Defendant admitted in his own statement not only that he pursued Sanders for his rings, but that he was the one who fired all the shots. This is not a case where "the precise content [or] even the existence of the [defendant’s] own confession were open to question”. (Cruz v New York, supra, 481 US, at 192; see, e.g., People v Galloway, 138 AD2d 735 [2d Dept 1988]; cf., People v Ray, 140 AD2d 380; People v Rodriguez, 138 AD2d 643; People v Latif, 135 AD2d 736, 738 [1987].)
Codefendant’s statement added the fact that defendant shot *429the deceased at close range in the chest. However, since the jury acquitted defendant of intentional murder, while convicting him of felony murder, this part of codefendant’s statement did not influence the jury’s verdict.
Defendant was identified in court by the token booth clerk.
Defendant’s primary defenses in this case were intoxication and renunciation. Far from undercutting these defenses, codefendant’s statements supplied support for defendant’s case.
There is thus no reasonable possibility that the jury would have acquitted defendant but for the admission of his codefendant’s statement into evidence, and the error was therefore harmless beyond a reasonable doubt.
Defendant’s motion for an order vacating the judgment of conviction against him pursuant to CPL 440.10 is therefore denied.
By "final”, the court meant cases in which a judgment of conviction had been rendered, the availability of appeal exhausted, and the time for petitioning for certiorari elapsed.