dissenting.
The Court today holds that no limitations need be placed on resentencings that occurred before the date of decision in North Carolina v. Pearce, 395 U. S. 711 (1969). I believe however, that the State has an obligation to present to the court reviewing the second conviction evidence from which that court can determine whether a new sentence, more severe than that imposed at a prior trial, resulted in part from the sentencing authority’s desire to punish the defendant for successfully appealing his first conviction.1 I therefore respectfully dissent.
I
This case raises the issue of retroactivity only because of the almost unbelievable sluggishness of the appellate process in Michigan. Payne’s second sentence was imposed on August 30, 1967, nearly two years before Pearce was decided. However, the Michigan Court of Appeals did not decide Payne’s appeal until July 28, 1969, one month after the decision in Pearce. The Michigan Supreme Court considered the case for two more years, finally deciding it on November 9, 1971. Had the appellate process in Michigan been at all expeditious, this Court might have used Payne’s case as the vehicle to de*60cide that harsher sentences on reconviction could be justified only by objective evidence of post-sentencing conduct by the defendant, the rule adopted in Pearce. The only difference between Pearce’s case and Payne’s, then, is that the former moved up to this Court more quickly than the latter. Different treatment of two cases is justified under our Constitution only when the cases differ in some respect relevant to the different treatment.2 And a difference in the speed with which a judicial system disposes of an appeal is not related in any way to the purposes served by the limitations that Pearce placed on resentencing. Thus, considerations of fairness rooted in the Constitution lead me to conclude that cases in the pipeline when a new constitutional rule is announced must be given the benefit of that rule.
The rule adopted by the Court today is curious in another way. The Court appears to say that a defendant who failed to appeal his first conviction out of “a reasonably based fear of actual vindictiveness,” ante, at 52 n. 5, is entitled to review of his conviction. Cf. Fay v. Noia, 372 U. S. 391, 396-397, n. 3 (1963).3 If his appeal is successful, his new trial will occur after the date of decision in Pearce. Thus, any new sentence will be *61subject to the limitations imposed by Pearce. The rather strange result is that someone like Payne, who adhered to state procedural rules for vindicating his right to an error-free trial, may receive an enhanced sentence without limitation, while someone who did not adhere to those rules may not have his sentence increased unless the requirements of Pearce are met. I suppose that anomalies are occasionally inevitable, but I submit that we should consider very carefully any rule of retro-activity that has the effect of penalizing compliance with state procedural rules.
II
The Court applies the now-familiar three-pronged test to determine whether Pearce should be given retroactive effect, and it reaches the now-familiar result of nonretro-activity.4 I believe that principled adjudication requires the Court to abandon the charade of carefully balancing countervailing considerations when deciding the question of retroactivity. Inspecting the cases dealing with retro-activity, I find that they appear to fall into three groups. In some cases, this Court has held that the trial court lacked jurisdiction in the traditional sense. See, e. g., Benton v. Maryland, 395 U. S. 784 (1969); Waller v. Florida, 397 U. S. 387 (1970). Those holdings have been made fully retroactive. Ashe v. Swenson, 397 U. S. 436 (1970); Robinson v. Neil, 409 U. S. 505 (1973). Cf. United States v. U. S. Coin & Currency, 401 U. S. 715 (1971). In other cases the Court announced a rule that was central to the process of determining guilt or innocence, and whose application might well have led to the *62acquittal of the defendant. See, e. g., Gideon v. Wainwright, 372 U. S. 335 (1963); In re Winship, 397 U. S. 358 (1970). Those holdings too have been given retroactive effect. Pickelsimer v. Wainwright, 375 U. S. 2 (1963); Ivan V. v. New York, 407 U. S. 203 (1972). Cf. Adams v. Illinois, 405 U. S. 278 (1972). All other constitutional rules of criminal procedure have been given prospective effect only.5
I confess that I have been unable to discover a principled basis for that threefold classification, but it does appear to be the factor operating in our cases. And I see little point in forcing lower courts to flounder without substantial guidance in the morass of our cases, by informing them that they are to apply a balancing test, when in fact it invariably occurs that the balancing test results in holdings of nonretroactivity. Furthermore, it demeans this Court to pretend to consider a variety of factors if, no matter how those factors are arrayed, the result is predetermined. An open-minded examination of this Court’s cases on retroactivity compels the conclusion that the Court divides cases into several classes, and it is the classification, not the three-pronged test, that determines the result. Our time would be better spent, I think, in attempting to delineate the basis for those classifications, and to derive them from some constitutional principles, rather than in “applying” a balancing test. Indeed, it might have been thought that *63Robinson v. Neil, supra, had begun the task of rationalizing our cases, but apparently that is not so.
Ill
The holding of Pearce is a simple one: the Due Process Clause requires States to adopt procedures designed to minimize the possibility that a new sentence after a successful appeal will be based in part on vindictiveness for the defendant’s having taken the appeal. The Court agrees that “this basic due process protection ... is available equally to defendants resentenced before and after the date of decision in that case.” Ante, at 50, 51. The question then is what procedures are required to insure that that protection has been afforded defendants re-sentenced before Pearce was decided. This question, like many of those involving retroactivity, relates to the integrity of the judicial process, not to the limitations placed by the Constitution on police behavior. One can agree that the precise requirements of Pearce are inappropriate for retrospective application, largely because they are procedurally ill-adapted to the problem, yet disagree with the Court that the States need do nothing at all to convince a reviewing court that vindictiveness played no part in the resentencing. See, e. g., Commonwealth v. Allen, 443 Pa. 96, 102, 277 A. 2d 803 (1971).
The issue need not be framed as the “retroactivity” of Pearce. The problem, as I see it, is to devise procedures that will permit reviewing courts to determine whether the requirements of the Due Process Clause have been met. In Pearce we concluded that it would be enough for a judge, on resentencing a defendant, to state his reasons for imposing a more severe sentence. If the more severe sentence was based upon objective information, placed on the record, concerning the conduct of the *64defendant after the first sentencing, the more severe sentence was permissible. Such a rule, although not absolutely guaranteeing that vindictiveness will play no part,6 nonetheless substantially reduces the possibility that it will, without significantly interfering with the judge’s lawful discretion.
A rather similar procedure would accomplish the same result for defendants resentenced before Pearce was decided. If a defendant did receive a harsher sentence after a successful appeal, and he seeks to have it reduced to the original sentence, the State should be required to present evidence that the new sentence was based on post-sentence conduct. In the absence of such evidence, the sentence must be reduced.7 The Court suggests that such a procedure would “occasion windfall benefits for some defendants who have suffered no constitutional deprivation.” Ante, at 53. That assertion must be considered more closely.
As the Court notes, there is little evidence that more severe sentences are often imposed. It cites an informal survey suggesting that 12% of reconvicted defendants receive higher sentences. Ante, at 56 n. 13. Even if that estimate is only half as large as the actual figure for pre-Pearce cases, still there are clearly very few defendants who have received harsher sentences. With respect *65to many of them, it will not be difficult to produce evidence supporting the new sentence. As in Moon v. Maryland, 398 U. S. 319 (1970), and Odom v. United States, 400 U. S. 23 (1970), the sentencing judge might indicate by affidavit or order the grounds for his sentencing decision. If memories have faded, the State might show that a presentence report considered by the judge recited post-sentence conduct by the defendant that would justify the harsher sentence.
Thus, I do not think that it can fairly be said that the requirements I would impose would in fact result in windfall benefits to “innumerable” defendants, ante, at 55; they would accrue to those few defendants who were convicted, successfully appealed, were reconvicted, and received harsher sentences so long ago that the State cannot produce evidence from which a reviewing court could find that vindictiveness played no part in the sentencing decision.8 And the “windfall benefits” would impair no substantial state interest in incarcerating those few offenders. Unlike the suppression of probative evidence that might severely limit the State's ability to secure a conviction of a person who undoubtedly committed an offense, here the remedy is simply the reduction of sentence. North Carolina v. Rice, 404 U. S. 244, 247 (1971). The sentence to be served would be one that had already been found appropriate by one *66judge, and would therefore satisfy the various interests advanced by incarceration.9
For these reasons, I dissent.
Mr. Justice Stewart joins Part III of this opinion.
The State did present an affidavit from the sentencing judge in this case. The Michigan Supreme Court held that it did not satisfy the requirement of North Carolina v. Pearce, 395 U. S. 711, 726 (1969), that more severe sentences can be justified only by “objective information concerning identifiable conduct on the part of the defendant occurring after the time of the original sentencing proceeding.” See 386 Mich. 84, 97, 191 N. W. 2d 375, 381 (1971). Petitioner contends that this holding was erroneous. Petition for Writ of Certiorari 5-6. The Court does not address this contention, nor shall I.
Since Payne’s appeal was pending when Pearce was decided, I need not consider whether different considerations, such as the defendant’s failure to raise the issue in seeking review from this Court or to persuade us on the merits, might suffice under the Due Process Clause to justify different treatment of defendants whose sentences had become final.
Mr. Justice Harlan, dissenting in Fay v. Noia, 372 U. S. 391, 475 (1963), suggested that the possibility of an enhanced sentence after a successful appeal, according to the Court, precluded the State from relying on a failure to appeal as an adequate state ground supporting the denial of relief under federal habeas corpus. On his interpretation, then, Fay anticipated the holding in Pearce.
In holding various rulings retroactive, this Court has given only the most cursory nod to the three-pronged test. See, e. g., Roberts v. Russell, 392 U. S. 293 (1968); McConnell v. Rhay, 393 U. S. 2 (1968); Arsenault v. Massachusetts, 393 U. S. 5 (1968).
Linkletter v. Walker, 381 U. S. 618 (1965), giving a limited retroactive effect to Mapp v. Ohio, 367 U. S. 643 (1961), is an anomaly at odds with the Court’s subsequent treatment of problems of retroactivity and can be explained only by the Court’s unfamiliarity with those problems when the case was decided. See also Johnson v. New Jersey, 384 U. S. 719 (1966).
For example, the sentencing judge, had he considered the case as an initial matter, might have imposed a sentence shorter than that imposed at the first trial, but, out of vindictiveness, he might decide to reimpose the original sentence. The procedures outlined in Pearce cannot prevent this.
I assume that the Court’s reliance on the continuing availability of the “foundational” right means that an offender who shows that vindictiveness played a part in his resentencing is entitled to relief. I would simply shift the burden of proof to the State, which has better access to the relevant facts.
State courts, closer to the problems of administering the rule I suggest, have widely thought that those burdens are not substantial. See, e. g., Stonom v. Wainwright, 235 So. 2d 545 (Fla. App. 1970); People v. Baze, 43 Ill. 2d 298, 253 N. E. 2d 392 (1969); State v. Pilcher, 171 N. W. 2d 251 (Iowa 1969); Hord v. Commonwealth, 450 S. W. 2d 530 (Ky. 1970); State v. Rentschler, 444 S. W. 2d 453 (Mo. 1969); Commonwealth v. Allen, 443 Pa. 96, 277 A. 2d 803 (1971); Denny v. State, 47 Wis. 2d 541, 178 N. W. 2d 38 (1970).
The Court’s conclusion that Pearce was not foreshadowed by decisions in this Court or by a trend of lower court decisions is somewhat misleading. This Court’s decision in Green v. United States, 355 U. S. 184 (1957), raised substantial questions under the Double Jeopardy Clause of the constitutionality of enhanced sentences after a successful appeal. Also, one reading of Fay v. Noia, 2Í12 U. S. 391 (1963), suggested by the dissent of Mr. Justice Harlan, is that a State may not burden the right to appeal with the possibility of an enhanced sentence. And prior to Pearce, the First, Second, Fourth, Fifth, and Seventh Circuits had held that enhanced sentences after reconviction could be justified only in limited circumstances. See Maraño v. United States, 374 F. 2d 583 (CA1 1967); United States v. Coke, 404 F. 2d 836 (CA2 1968) (en banc); Patton v. North Carolina, 381 F. 2d 636 (CA4 1967); Simpson v. Rice, 396 F. 2d 499 (CA5 1968); United States v. White, 382 F. 2d 445 (CA7 1967). So had the California Supreme Court, in a powerful opinion by Justice Traynor. People v. Henderson, 60 Cal. 2d 482, 386 P. 2d 677 (1963). Finally, a “learned and effective article,” as Judge Friendly called it in United States v. Coke, supra, arguing the same point, appeared in 1965. Van Alstyne, In Gideon’s Wake: Harsher Penalties and the “Successful” Criminal Appellant, 74 Yale L. J. 606 (1965). I would think that these decisions and commentary had prepared the ground rather well for Pearce, as the Court concedes, ante, at 56 n. 12. Yet if the result was foreshadowed, it is not unreasonable to require States now to supplement the record, so that it will be clear that unconstitutional sentences were not imposed. Because it insists on treating the issue here as a question of retroactivity, the Court does not address this argument.