delivered the opinion of the Court.
In Payton v. New York, 445 U. S. 573 (1980), this Court held that the Fourth Amendment1 prohibits the police from making a warrantless and nonconsensual entry into a sus*539pect’s home to make a routine felony arrest. The question before us in the present case is whether the rule announced in Payton applies to an arrest that took place before Payton was decided.
I
Special Agents Hemenway and Pickering of the United States Secret Service suspected respondent Raymond Eugene Johnson and his codefendant, Oscar Joseph Dodd, of attempting to negotiate a misdelivered United States Tréasury check.2 Proceeding without an arrest warrant, on May 5, 1977, the two agents went to respondent’s Los Angeles home and waited outside. Shortly thereafter, respondent and his wife arrived and entered the house.
The agents drew their weapons, approached the doorway and knocked, identifying themselves by fictitious names. When respondent opened the door, he saw the two agents with their guns drawn and their badges raised. Respondent permitted the agents to enter the house. While one agent stood with respondent in the living room, the other searched the premises. The agents then advised respondent of his constitutional rights and interrogated him. When respondent revealed his involvement in the taking of the misdeliv-ered check, the agents formally arrested him. Respondent later signed a written statement admitting his involvement with the check.
Before trial, respondent sought to suppress his oral and written statements as fruits of an unlawful arrest not sup*540ported by probable cause. The United States District Court for the Central District of California found respondent’s arrest to be proper and admitted the evidence. App. 7. A jury then convicted respondent of aiding and abetting obstruction of correspondence, in violation of 18 U. S. C. §§2 and 1702.3 The imposition of respondent’s sentence was suspended in favor of five years’ probation.
By an unreported opinion filed December 19, 1978, the United States Court of Appeals for the Ninth Circuit affirmed the judgment of conviction. Acknowledging that “[i]t certainly would have been preferable had the agents obtained a warrant” for respondent’s arrest before entering his residence, the court nonetheless ruled that “if probable cause exists for the arrest, [respondent’s] constitutional rights were not violated by the warrantless arrest, even though there may have been time [for the agents] to have obtained a warrant for his arrest.” App. to Pet. for Cert. 26a-27a.
On April 15, 1980, while respondent’s petition for rehearing was still pending before the Ninth Circuit, this Court decided Payton v. New York, supra.4 On September 2, *5411980, the Ninth Circuit granted respondent’s petition for rehearing, withdrew its prior opinion, and on the strength of Payton, now reversed the judgment of conviction. 626 F. 2d 753. “In light of the strong language by the Court in Payton emphasizing the special protection the Constitution affords to individuals within their homes,” the Court of Appeals held that “the warrantless arrest of Johnson, while he stood within his home, after having opened the door in response to false identification by the agents, constituted a violation of his Fourth Amendment rights.” Id., at 757. The Government petitioned for rehearing, arguing that the principles of Payton should not apply retroactively to an arrest that had occurred before Payton was decided. The Court of Appeals disagreed, denied the petition for rehearing, and amended its opinion to clarify that Payton did apply retroactively. App. to Pet. for Cert. 12a.5
The Government sought review in this Court. We granted certiorari to consider the retrospective effect, if any, of the Fourth Amendment rule announced in Payton. 454 U. S. 814 (1981).6
*542II
“[T]he federal constitution has no voice upon the subject” of retrospectivity. Great Northern R. Co. v. Sunburst Oil & Refining Co., 287 U. S. 358, 364 (1932). Before 1965, when this Court decided Linkletter v. Walker, 381 U. S. 618, “both the common law and our own decisions recognized a general rule of retrospective effect for the constitutional decisions of this Court. . . subject to [certain] limited exceptions.” Robinson v. Neil, 409 U. S. 505, 507 (1973), citing Norton v. Shelby County, 118 U. S. 425, 442 (1886), and Chicot County Drainage Dist. v. Baxter State Bank, 308 U. S. 371 (1940).7
In Linkletter, however, the Court concluded “that the Constitution neither prohibits nor requires [that] retrospective effect” be given to any “new” constitutional rule. 381 U. S., at 629. Since Linkletter, the Court’s announcement of a constitutional rule in the realm of criminal procedure has frequently been followed by a separate decision explaining whether, and to what extent, that rule applies to past, pending, and future cases. See generally Beytagh, Ten Years of Non-Retroactivity: A Critique and a Proposal, 61 Va. L. Rev. 1557 (1975).
Linkletter itself addressed the question whether the Fourth Amendment exclusionary rule of Mapp v. Ohio, 367 U. S. 643 (1961), should apply to state convictions that had become final before Mapp was decided.8 At the outset, the Linkletter Court noted that cases still pending on direct review when Mapp was handed down had already received the *543benefit of Mapp’s rule. See 381 U. S., at 622, n. 4, citing Ker v. California, 374 U. S. 23 (1963); Fahy v. Connecticut, 375 U. S. 85 (1963); and Stoner v. California, 376 U. S. 483 (1964). This limited retrospective application of Mapp was consistent with the common-law rule, recognized in both civil and criminal litigation, “that a change in law will be given effect while a case is on direct review.” 381 U. S., at 627, citing United States v. Schooner Peggy, 1 Cranch 103 (1801).
To determine whether a particular ruling should also extend to cases that were already final, Linkletter directed courts to “weigh the merits and demerits in each case by looking to the prior history of the rule in question, its purpose and effect, and whether retrospective operation will further or retard its operation.” 381 U. S., at 629. Employing that test, the Court concluded that the Mapp rule should not apply to convictions that had become final before Mapp was decided.
The following Term, in Tehan v. United States ex rel. Shott, 382 U. S. 406 (1966), the Court applied Linkletter’s analysis to hold the Fifth Amendment rule of Griffin v. California, 380 U. S. 609 (1965) (barring comment on a state defendant’s failure to testify), nonretroactive to judgments of conviction made final before Griffin was decided. The Court again found no “question of the applicability of the Griffin rule to cases still pending on direct review at the time it was announced.” 382 U. S., at 409, n. 3, citing O’Connor v. Ohio, 382 U. S. 286 (1965). Thus, after Linkletter and Shott, it appeared that all newly declared constitutional rules of criminal procedure would apply retrospectively at least to judgments of conviction not yet final when the rule was established.
In Johnson v. New Jersey, 384 U. S. 719 (1966), and Stovall v. Denno, 388 U. S. 293 (1967), however, the Court departed from that basic principle. Those cases held that, in the interest of justice, the Court may balance three factors to determine whether a “new” constitutional rule should be ret-*544respectively or prospeetively applied: “(a) the purpose to be served by the new standards, (b) the extent of the reliance by law enforcement authorities on the old standards, and (c) the effect on the administration of justice of a retroactive application of the new standards.” Id., at 297. See also Johnson v. New Jersey, 384 U. S., at 728. Because the outcome of that balancing process might call for different degrees of retroactivity in different cases, the Court concluded that “no distinction is justified between convictions now final . . . and convictions at various stages of trial and direct review.” Stovall v. Denno, 388 U. S., at 300. See Johnson v. New Jersey, 384 U. S., at 732.
Because the balance of the three Stovall factors inevitably has shifted from case to case, it is hardly surprising that, for some, “the subsequent course of Linkletter became almost as difficult to follow as the tracks made by a beast of prey in search of its intended victim.” Mackey v. United States, 401 U. S. 667, 676 (1971) (separate opinion of Harlan, J.). At one extreme, the Court has regularly given complete retroactive effect to new constitutional rules whose major purpose “is to overcome an aspect of the criminal trial that substantially impairs its truth-finding function and so raises serious questions about the accuracy of guilty verdicts in past trials.” Williams v. United States, 401 U. S. 646, 653 (1971) (plurality opinion). See also id., at 653, n. 6; Brown v. Louisiana, 447 U. S. 323, 328-330 (1980) (plurality opinion); Hankerson v. North Carolina, 432 U. S. 233, 243 (1977); Gosa v. Mayden, 413 U. S. 665, 679 (1973) (plurality opinion); Ivan V. v. City of New York, 407 U. S. 203, 205 (1972).
At the other extreme, the Court has applied some standards only to future cases, denying the benefit of the new rule even to the parties before the Court. See, e. g., Morrissey v. Brewer, 408 U. S. 471, 490 (1972) (establishing basic requirements applicable only to “future revocations of parole”). Cf. Johnson v. New Jersey, 384 U. S., at 733, citing England *545v. Louisiana State Board of Medical Examiners, 375 U. S. 411 (1964), and James v. United States, 366 U. S. 213 (1961). As an intermediate position, the Court has applied a change in the law to all future litigants, but retroactively only to the parties at bar. See, e. g., Stovall v. Denno, 388 U. S., at 301; DeStefano v. Woods, 392 U. S. 631, 633 (1968); Adams v. Illinois, 405 U. S. 278, 284-285 (1972) (plurality opinion); Michigan v. Payne, 412 U. S. 47 (1973).
In a consistent stream of separate opinions since Link-letter, Members of this Court have argued against selective awards of retroactivity. Those opinions uniformly have asserted that, at a minimum, all defendants whose cases were still pending on direct appeal at the time of the law-changing decision should be entitled to invoke the new rule.9 *546In Desist v. United States, 394 U. S. 244, 256 (1969) (dissenting opinion), and Mackey v. United States, 401 U. S., at 675 (separate opinion), Justice Harlan presented a comprehensive analysis in support of that principle. In his view, failure to apply a newly declared constitutional rule at least to cases pending on direct review at the time of the decision violated three norms of constitutional adjudication.
First, Justice Harlan argued, the Court’s “ambulatory ret-roactivity doctrine,” id., at 681, conflicts with the norm of principled decisionmaking. “Some members of the Court, and I have come to regret that I was among them, initially grasped this doctrine as a way of limiting the reach of decisions that seemed to them fundamentally unsound. Others rationalized this resort to prospectivity as a ‘technique’ that provided an ‘impetus ... for the implementation of long overdue reforms, which otherwise could not be practicably effected.’” Id., at 676, citing Jenkins v. Delaware, 395 U. S. 213, 218 (1969). “The upshot of this confluence of viewpoints,” 401 U. S., at 676, was that the coalitions favoring nonretroactivity had realigned from case to case, inevitably generating a welter of “incompatible rules and inconsistent principles,” Desist v. United States, 394 U. S., at 258. See also Michigan v. Payne, 412 U. S., at 61 (Marshall, J., dissenting) (“principled adjudication requires the Court to abandon the charade of carefully balancing countervailing considerations when deciding the question of retroactivity”).
Second, Justice Harlan found it difficult to accept the notion that the Court, as a judicial body, could apply a “ ‘new* constitutional rule entirely prospectively, while making an exception only for the particular litigant whose case was chosen as the vehicle for establishing that rule.” Desist v. *547United. States, 394 U. S., at 258 (dissenting opinion). A legislature makes its new rules “wholly or partially retroactive or only prospective as it deems wise.” Mackey v. United States, 401 U. S., at 677 (Harlan, J., dissenting). This Court, however,
“announcefs] new constitutional rules . . . only as a correlative of our dual duty to decide those cases over which we have jurisdiction and to apply the Federal Constitution as one source of the matrix of governing legal rules. . . . Simply fishing one case from the stream of appellate review, using it as a vehicle for pronouncing new constitutional standards, and then permitting a stream of similar cases subsequently to flow by unaffected by that new rule constitute an indefensible departure from this model of judicial review.” Id., at 678-679.
Third, Justice Harlan asserted that the Court’s selective application of new constitutional rules departed from the principle of treating similarly situated defendants similarly:10
“[W]hen another similarly situated defendant comes before us, we must grant the same relief or give a principled reason for acting differently. We depart from this basic judicial tradition when we simply pick and choose from among similarly situated defendants those who *548alone will receive the benefit of a ‘new’ rule of constitutional law.” Desist v. United States, 394 U. S., at 258-259 (dissenting opinion).
Justice Harlan suggested one simple rule to satisfy all three of his concerns. “I have concluded that Linkletter was right in insisting that all ‘new’ rules of constitutional law must, at a minimum, be applied to all those cases which are still subject to direct review by this Court at the time the ‘new’ decision is handed down.” Id., at 258. “[A] proper perception of our duties as a court of law, charged with applying the Constitution to resolve every legal dispute within our jurisdiction on direct review, mandates that we apply the law as it is at the time, not as it once was.” Mackey v. United States, 401 U. S., at 681 (separate opinion).
We now agree with Justice Harlan that “‘[rjetroactivity’ must be rethought,” Desist v. United States, 394 U. S., at 258 (dissenting opinion). We therefore examine the circumstances of this case to determine whether it presents a retro-activity question clearly controlled by past precedents, and if not, whether application of the Harlan approach would resolve the retroactivity issue presented in a principled and equitable manner.
III
A
At the outset, we must first ask whether respondent’s case presents a retrospectivity problem clearly controlled by existing precedent. Re-examination of the post -Linkletter decisions convinces us that in three narrow categories of cases, the answer to the retroactivity question has been effectively determined, not by application of the Stovall factors, but rather, through application of a threshold test.11
*549First, when a decision of this Court merely has applied settled precedents to new and different factual situations, no real question has arisen as to whether the later decision should apply retrospectively. In such cases, it has been a foregone conclusion that the rule of the later case applies in earlier cases, because the later decision has not in fact altered that rule in any material way. See, e. g., Dunaway v. New York, 442 U. S. 200, 206 (1979) (reviewing application of the rule in Brown v. Illinois, 422 U. S. 590 (1975)); Spinelli v. United States, 393 U. S. 410, 412 (1969) (“further explicat[ing]” the principles of Aguilar v. Texas, 378 U. S. 108 (1964)); Desist v. United States, 394 U. S., at 263 (Harlan, J., dissenting).
Conversely, where the Court has expressly declared a rule of criminal procedure to be “a clear break with the past,” Desist v. United States, 394 U. S., at 248, it almost invariably has gone on to find such a newly minted principle nonretroac-tive. See United States v. Peltier, 422 U. S. 531, 547, n. 5 (1975) (Brennan, J., dissenting) (collecting cases). In this second type of case, the traits of the particular constitutional rule have been less critical than the Court’s express threshold determination that the “‘new’ constitutional interpreta-tio[n]... so change[s] the law that prospectivity is arguably the proper course,” Williams v. United States, 401 U. S., at 659 (plurality opinion). Once the Court has found that the new rule was unanticipated, the second and third Stovall factors — reliance by law enforcement authorities on the old standards and effect on the administration of justice of a retroactive application of the new rule — have virtually *550compelled a finding of nonretroactivity. See, e. g., Gosa v. Mayden, 413 U. S., at 672-673, 682-685 (plurality opinion); Michigan v. Payne, 412 U. S., at 55-57.12
Third, the Court has recognized full retroactivity as a necessary adjunct to a ruling that a trial court lacked authority to convict or punish a criminal defendant in the first place. The Court has invalidated inconsistent prior judgments where its reading of a particular constitutional guarantee immunizes a defendant’s conduct from punishment, see, e. g., United States v. United States Coin & Currency, 401 U. S. 715, 724 (1971) (penalty against assertion of Fifth Amendment privilege against self-incrimination), or serves “to prevent [his] trial from taking place at all, rather than to prescribe procedural rules that govern the conduct of [that] trial,” Robinson v. Neil, 409 U. S., at 509 (double jeopardy). In such cases, the Court has relied less on the technique of retroactive application than on the notion that the prior inconsistent judgments or sentences were void ab initio. See, e. g., Moore v. Illinois, 408 U. S. 786, 800 (1972) (retroactive application of Eighth Amendment ruling in Furman v. Georgia, 408 U. S. 238 (1972)); Ashe v. Swenson, 397 U. S. 436, 437, n. 1 (1970) (retroactive application of double jeopardy ruling in Benton v. Maryland, 395 U. S. 784 (1969)). See also Gosa v. Mayden, 413 U. S., at 693 (Marshall, J., dissenting); Michigan v. Payne, 412 U. S., at 61 (Marshall, J., dissenting) (rulings are fully retroactive when the “Court *551has held that the trial court lacked jurisdiction in the traditional sense”).
Respondent’s case neatly fits none of these three categories. First, Payton v. New York did not simply apply settled precedent to a new set of facts. In Payton, the Court acknowledged that the “important constitutional question presented” there had been “expressly left open in a number of our prior opinions.” 445 U. S., at 574 and 575, n. 1, citing United States v. Watson, 423 U. S. 411, 418, n. 6 (1976); Gerstein v. Pugh, 420 U. S. 103, 113, n. 13 (1975); Coolidge v. New Hampshire, 403 U. S. 443, 474-481 (1971); and Jones v. United States, 357 U. S. 493, 499-500 (1958).
By the same token, however, Payton also did not announce an entirely new and unanticipated principle of law. In general, the Court has not subsequently read a decision to work a “sharp break in the web of the law,” Milton v. Wainwright, 407 U. S. 371, 381, n. 2 (1972) (Stewart, J., dissenting), unless thai ruling caused “such an abrupt and fundamental shift in doctrine as to constitute an entirely new rule which in effect replaced an older one,” Hanover Shoe, Inc. v. United Shoe Machinery Corp., 392 U. S. 481, 498 (1968). Such a break has been recognized only when a decision explicitly overrules a past precedent of this Court, see, e. g., Desist v. United States, 394 U. S. 244 (1969); Williams v. United States, 401 U. S. 646 (1971), or disapproves a practice this Court arguably has sanctioned in prior cases, see, e. g., Gosa v. Mayden, 413 U. S., at 673 (plurality opinion); Adams v. Illinois, 405 U. S., at 283; Johnson v. New Jersey, 384 U. S., at 731, or overturns a longstanding and widespread practice to which this Court has not spoken, but which a near-unanimous body of lower court authority has expressly approved. See, e. g., Gosa v. Mayden, 413 U. S., at 673 (plurality opinion) (applying nonretroactively a decision that “effected a decisional change in attitude that had prevailed for many decades”); Stovall v. Denno, 388 U. S., at 299-300. See also Chevron Oil Co. v. Huson, 404 U. S. 97, 107 (1971); Cipriano *552v. City of Houma, 395 U. S. 701 (1969); Milton v. Wainwright, 407 U. S., at 381-382, n. 2 (Stewart, J., dissenting) (“sharp break” occurs when “decision overrules clear past precedent ... or disrupts a practice long accepted and widely relied upon”).
Payton did none of these. Payton expressly overruled no clear past precedent of this Court on which litigants may have relied. Nor did Payton disapprove an established practice that the Court had previously sanctioned. To the extent that the Court earlier had spoken to the conduct engaged in by the police officers in Payton, it had deemed it of doubtful constitutionality.13 The Court’s own analysis in Payton makes it clear that its ruling rested on both long-recognized principles of Fourth Amendment law and the weight of historical authoiity as it had appeared to the Framers of the Fourth Amendment.14 Finally, Payton overturned no long*553standing practice approved by a near-unanimous body of lower court authority.15 Payton therefore does not fall into that narrow class of decisions whose nonretroactivity is effec*554tively preordained because they unmistakably signal “a clear break with the past,” Desist v. United States, 394 U. S., at 248.
It is equally plain that Payton does not fall into the third category of cases that do not pose difficult retroactivity questions. Payton did not hold that the trial court lacked authority to convict or sentence Theodore Payton, nor did Payton’s reading of the Fourth Amendment immunize Payton’s conduct from punishment. The holding in Payton did not prevent the defendant’s trial from taking place; rather, it reversed the New York Court of Appeals’ judgment and remanded for a new trial to be conducted without unconstitutionally obtained evidence.
B
Having determined that the retroactivity question here is not clearly controlled by our prior precedents, we next must ask whether that question would be fairly resolved by applying the rule in Payton to all cases still pending on direct appeal at the time when Payton was decided. Answering that question affirmatively would satisfy each of the three concerns stated in Justice Harlan’s opinions in Desist and Mackey.
First, retroactive application of Payton to all previously nonfinal convictions would provide a principle of decisionmak-ing consonant with our original understanding of retroactivity in Linkletter and Shott. Moreover, such a principle would be one capable of general applicability, satisfying Justice Harlan’s central concern: “Refusal to apply new constitutional rules to all cases arising on direct review . . . tends to cut this Court loose from the force of precedent, allowing us *555to restructure artificially those expectations legitimately created by extant law and thereby mitigate the practical force of stare decisis ... a force which ought properly to bear on the judicial resolution of any legal problem.” Mackey v. United States, 401 U. S., at 680-681 (separate opinion).
Second, application of Payton to cases pending on direct review would comport with our judicial responsibilities “to do justice to each litigant on the merits of his own case,” Desist v. United States, 394 U. S., at 259 (Harlan, J., dissenting), and to “resolve all cases before us on direct review in light of our best understanding of governing constitutional principles.” Mackey v. United States, 401 U. S., at 679 (separate opinion of Harlan, J.). The Court of Appeals held that the circumstances of respondent’s arrest violated Payton, and the Government does not dispute that contention. See n. 6, supra. It would be ironic indeed were we now to reverse a judgment applying Payton’s rule, when in Payton itself, we reversed a directly contrary judgment of the New York Court of Appeals. As Justice Harlan noted in Desist: “If a ‘new’ constitutional doctrine is truly right, we should not reverse lower courts which have accepted it; nor should we affirm those which have rejected the very arguments we have embraced.” 394 U. S., at 259.
Third, application of the Harlan approach to respondent’s case would further the goal of treating similarly situated defendants similarly. The Government contends that respondent may not invoke Payton because he was arrested before Payton was decided. Yet it goes without saying that Theodore Payton also was arrested before Payton was decided, and he received the benefit of the rule in his case. Furthermore, at least one other defendant whose conviction was not final when Payton issued benefited from Payton’s rule, although he, too, was arrested before Payton was decided.16 *556An approach that resolved all nonfinal convictions under the same rule of law would lessen the possibility that this Court might mete out different constitutional protection to defendants simultaneously subjected to identical police conduct.17
*557IV
Against adoption of this approach, the Government raises four arguments based on United States v. Peltier, 422 U. S. 531 (1975). None is persuasive.
The Government first cites Peltier’s, holding: that the Fourth Amendment rule announced in Almeida-Sanchez v. United States, 413 U. S. 266 (1973), should not apply retroactively to a case pending on appeal when Almeida-Sanchez was announced. By so holding, the Government suggests, *558Peltier declared a principle that controls the issue of retro-activity for all Fourth Amendment rulings.18
Upon examination, however, the retroactivity question posed here differs from that presented in Peltier. As the Government concedes, Payton overturned neither a statute nor any consistent judicial history approving noneonsensual, warrantless home entries. See Brief for United States 30, n. 18. Thus, its nonretroactivity is not preordained under the “clear break” principles stated above. In Peltier, in contrast, the Court noted that Almeida-Sanchez had invalidated a form of search previously sanctioned by “a validly enacted statute, supported by longstanding administrative regulations and continuous judicial approval.” 422 U. S., at 541. See also Almeida-Sanchez v. United States, 413 U. S., at 278 (Powell, J., concurring) (“While the question is one of first impression in this Court,” the practice disapproved had “been consistently approved by the judiciary”); id., at 298-299, n. 10 (White, J., dissenting) (35 of 36 judges in 20 Court of Appeals cases had approved the invalidated practice).
Because Almeida-Sanchez had overturned a longstanding practice to which this Court had not spoken, but which a near-unanimous body of lower court authority had approved, it represented a “clear break” with the past. For that reason alone, under controlling retroactivity precedents, the nonretroactive application of Almeida-Sanchez would have been appropriate even if the case had involved no Fourth Amendment question. In that respect, Peltier resembles several earlier decisions that held “new” Fourth Amendment *559doctrine nonretroactive, not on the ground that all Fourth Amendment rulings apply only prospectively, but because the particular decisions being applied “so change[d] the law that prospectivity [was] arguably the proper course.” Williams v. United States, 401 U. S., at 659 (plurality opinion) (refusing to apply retroactively Chimel v. California, 395 U. S. 752 (1969), which overruled United States v. Rabinowitz, 339 U. S. 56 (1950), and Harris v. United States, 331 U. S. 145 (1947)). See also Desist v. United States, 394 U. S. 244 (1969) (refusing to apply retroactively Katz v. United States, 389 U. S. 347 (1967), which overruled Goldman v. United States, 316 U. S. 129 (1942), and Olmstead v. United States, 277 U. S. 438 (1928)).
The Government bases its second argument on Peltier’s, broad language: “If the purpose of the exclusionary rule is to deter unlawful police conduct then evidence obtained from a search should be suppressed only if it can be said that the law enforcement officer had knowledge, or may properly be charged with knowledge, that the search was unconstitutional under the Fourth Amendment” (emphasis added). 422 U. S., at 542. The Government reads this language to require that new Fourth Amendment rules must be denied retroactive effect in all cases except those in which law enforcement officers failed to act in good-faith compliance with then-prevailing constitutional norms.
The Government does not seriously suggest that the retro-activity of a given Fourth Amendment ruling should turn solely on the subjective state of a particular arresting officer’s mind. Instead, it offers an “objective” test: that law enforcement officers “may properly be charged with knowledge” of all “settled” Fourth Amendment law. Under the Government’s theory, because the state of Fourth Amendment law regarding warrantless home arrests was “unsettled” before Payton, that ruling should not apply retroactively even to cases pending on direct appeal when Payton was decided. See Brief for United States 14-19, 34-38.
*560Yet the Government’s reading of Peltier would reduce its own “retroactivity test” to an absurdity. Under this view, the only Fourth Amendment rulings worthy of retroactive application are those in which the arresting officers violated pre-existing guidelines clearly established by prior cases. But as we have seen above, cases involving simple application of clear, pre-existing Fourth Amendment guidelines raise no real questions of retroactivity at all. Literally read, the Government’s theory would automatically eliminate all Fourth Amendment rulings from consideration for retroactive application.
The Government’s third claim is that Peltier’s logic suggests that retroactive application of Fourth Amendment decisions like Payton — even to cases pending on direct review— would not serve the policies underlying the exclusionary rule. Cf. 422 U. S., at 536-542. Yet viewed in the light of Peltier’s holding, this assertion also fails. Peltier suggested only that retroactive application of a Fourth Amendment ruling that worked a “sharp break” in the law, like Almeida-Sanchez, would have little deterrent effect, because law enforcement officers would rarely be deterred from engaging in a practice they never expected to be invalidated. See 422 U. S., at 541-542.
This logic does not apply to a ruling like Payton, that resolved a previously unsettled point of Fourth Amendment law. Because this Court cannot rule on every unsettled Fourth Amendment question, years may pass before the Court finally invalidates a police practice of dubious constitutionality. See, e. g., Desist v. United States, 394 U. S., at 275 (Fortas, J., dissenting) (arguing that the “physical-trespass” wiretap rule of Olmstead v. United States, 277 U. S. 438 (1928), had been moribund for 17 years before it was formally overruled). Long before Payton, for example, this Court had questioned the constitutionality of warrantless home arrests. See n. 13, supra. Furthermore, the Court’s *561opinions consistently had emphasized that, in light of the constitutional protection traditionally accorded to the privacy of the home, police officers should resolve any doubts regarding the validity of a home arrest in favor of obtaining a warrant. See, e. g., Johnson v. United States, 333 U. S. 10, 14 (1948) (“Any assumption that evidence sufficient to support a magistrate’s disinterested determination to issue a search warrant will justify the officers in making a search without a warrant would reduce the Amendment to a nullity and leave the péo-ple’s homes secure only in the discretion of police officers”).
If, as the Government argues, all rulings resolving unsettled Fourth Amendment questions should be nonretroactive, then, in close cases, law enforcement officials would have little incentive to err on the side of constitutional behavior.19 Official awareness of the dubious constitutionality of a practice would be counterbalanced by official certainty that, so long as the Fourth Amendment law in the area remained unsettled, evidence obtained through the questionable practice would be excluded only in the one case definitively resolving the unsettled question. Failure to accord any retroactive effect to Fourth Amendment rulings would “encourage police or other courts to disregard the plain purport of our decisions and to adopt a let’s-wait-until-it’s-decided approach.” Desist v. United States, 394 U. S., at 277 (Fortas, J., dissenting).
The Government finally argues that retroactive application of Payton, even to a case pending on direct appeal, would accomplish nothing but the discharge of a wrongdoer. Justice Harlan gave the answer to this assertion. “We do not release a criminal from jail because we like to do so, or because we think it wise to do so, but only because the government has offended constitutional principle in the conduct of his case. And when another similarly situated defendant comes *562before us, we must grant the same relief or give a principled reason for acting differently.” Desist v. United States, 394 U. S., at 258 (dissenting opinion). Applying Payton to convictions that were not yet final when Payton issued would accomplish the first step toward “turning our backs on the ad hoc approach that has so far characterized our decisions in the retroactivity field and proceeding to administer the doctrine on principle.” Jenkins v. Delaware, 395 U. S., at 224 (Harlan, J., dissenting).
V
To the extent necessary to decide today’s case, we embrace Justice Harlan’s views in Desist and Mackey. We therefore hold that, subject to the exceptions stated below, a decision of this Court construing the Fourth Amendment is to be applied retroactively to all convictions that were not yet final at the time the decision was rendered.
By so holding, however, we leave undisturbed our precedents in other areas. First, our decision today does not affect those cases that would be clearly controlled by our existing retroactivity precedents. Second, because respondent’s case arises on direct review, we need not address the retroactive reach of our Fourth Amendment decisions to those cases that still may raise Fourth Amendment issues on collateral attack.20 Cf. n. 10, supra. Third, we express no view on the retroactive application of decisions construing any constitutional provision other than the Fourth Amendment.21 *563Finally, all questions of civil retroactivity continue to be governed by the standard enunciated in Chevron Oil Co. v. Huson, 404 U. S., at 106-107. See n. 12, supra.
Respondent’s case was pending on direct appeal when Payton v. New York was decided. Because the Court of Appeals correctly held that the rule in Payton should apply to respondent’s case, its judgment is affirmed.22
It is so ordered.