*97Mr. Justice Stewart

delivered the opinion of the Court.

The respondent, Richard Bert Mosley, was arrested in Detroit, Mich., in the early afternoon of April 8, 1971, in connection with robberies that had recently occurred at the Blue Goose Bar and the White Tower Restaurant on that city’s lower east side. The arresting officer, Detective James Cowie of the Armed Robbery Section of the Detroit Police Department, was acting on a tip implicating Mosley and three other men in the robberies.1 After effecting the arrest, Detective Cowie brought Mosley to the Robbery, Breaking and Entering Bureau of the Police Department, located on the fourth floor of the departmental headquarters building. The officer advised Mosley of his rights under this Court’s decision in Miranda v. Arizona, 384 U. S. 436, and had him read and sign the department’s constitutional rights notification certificate. After filling out the necessary arrest papers, Cowie began questioning Mosley about the robbery of the White Tower Restaurant. When Mosley said he did not want to answer any questions about the robberies, Cowie promptly ceased the interrogation. The completion of the arrest papers and the questioning of Mosley together took approximately 20 minutes. At no time during the questioning did Mosley indicate a desire to consult with a lawyer, and there is no claim that the procedures followed to this point did not fully comply with the strictures of the Miranda opinion. Mosley was then taken to a ninth-floor cell block.

Shortly after 6 p. m., Detective Hill of the Detroit *98Police Department Homicide Bureau brought Mosley from the cell block to the fifth-floor office of the Homicide Bureau for questioning about the fatal shooting of a man named Leroy Williams. Williams had been killed on January 9, 1971, during a holdup attempt outside the 101 Ranch Bar in Detroit. Mosley had not been arrested on this charge or interrogated about it by Detective Cowie.2 Before questioning Mosley about this homicide, Detective Hill carefully advised him of his “Miranda rights.” Mosley read the notification form both silently and aloud, and Detective Hill then read and explained the warnings to him and had him sign the form. Mosley at first denied any involvement in the Williams murder, but after the officer told him that Anthony Smith had confessed to participating in the slaying and had named him as the “shooter,” Mosley made a statement impheating himself in the homicide.3 The interrogation by Detective Hill lasted approximately 15 minutes, and at no time during its course did Mosley ask to consult with a lawyer or indicate that he did not want to discuss the homicide. In short, there is no claim that the procedures followed during Detective Hill’s interrogation of Mosley, standing alone, did not fully comply with the strictures of the Miranda opinion.4

Mosley was subsequently charged in a one-count information with first-degree murder. Before the trial he moved to suppress his incriminating statement on a number of grounds, among them the claim that under the doctrine of the Miranda case it was constitutionally *99impermissible for Detective Hill to question him about the Williams murder after he had told Detective Cowie that he did not want to answer any questions about the robberies.5 The trial court denied the motion to suppress after an evidentiary hearing, and the incriminating statement was subsequently introduced in evidence against Mosley at his trial. The jury convicted Mosley of first-degree murder, and the court imposed a mandatory sentence of life imprisonment.

On appeal to the Michigan Court of Appeals, Mosley renewed his previous objections to the use of his incriminating statement in evidence. The appellate court reversed the judgment of conviction, holding that Detective Hill’s interrogation of Mosley had been a per se violation of the Miranda doctrine. Accordingly, without reaching Mosley’s other contentions, the Court remanded the case for a new trial with instructions that Mosley’s statement be suppressed as evidence. 51 Mich. App. 105, 214 N. W. 2d 564. After further appeal was denied by the Michigan Supreme Court, 392 Mich. 764, the State filed a petition for certiorari here. We granted the writ because of the important constitutional question presented. 419 U. S. 1119.

In the Miranda case this Court promulgated a set of safeguards to protect the there-delineated constitutional rights of persons subjected to custodial police interrogation. In sum, the Court held in that case that unless law enforcement officers give certain specified warnings be*100fore questioning a person in custody,6 and follow certain specified procedures during the course of any subsequent interrogation, any statement made by the person in custody cannot over his objection be admitted in evidence against him as a defendant at trial, even though the statement may in fact be wholly voluntary. See Michigan v. Tucker, 417 U. S. 433, 443.

Neither party in the present case challenges the continuing validity of the Miranda decision, or of any of the so-called guidelines it established to protect what the Court there said was a person’s constitutional privilege against compulsory self-incrimination. The issue in this case, rather, is whether the conduct of the Detroit police that led to Mosley’s incriminating statement did in fact violate the Miranda “guidelines,” so as to render the statement inadmissible in evidence against Mosley at his trial. Resolution of the question turns almost entirely on the interpretation of a single passage in the Miranda opinion, upon which the Michigan appellate court relied in finding a per se violation of Miranda:

“Once warnings have been given, the subsequent procedure is clear. If the individual indicates in any manner, at any time prior to or during questioning, that he wishes to remain silent, the interrogation must cease. At this point he has shown that he intends to exercise his Fifth Amendment privilege; any statement taken after the person invokes his privilege cannot be other than the product of compulsion, subtle or otherwise. Without the right to cut off questioning, the setting of in-custody *101interrogation operates on the individual to overcome free choice in producing a statement after the privilege has been once invoked.” 384 U. S., at 473-474.7

This passage states that “the interrogation must cease” when the person in custody indicates that “he wishes to remain silent.” It does not state under what circumstances, if any, a resumption of questioning is permissible.8 The passage could be literally read to mean that *102a person who has invoked his “right to silence” can never again be subjected to custodial interrogation by any police officer at any time or place on any subject. Another possible construction of the passage would characterize “any statement taken after the person invokes his privilege” as “the product of compulsion” and would therefore mandate its exclusion from evidence, even if it were volunteered by the person in custody without any further interrogation whatever. Or the passage could be interpreted to require only the immediate cessation of questioning, and to permit a resumption of interrogation after a momentary respite.

It is evident that any of these possible literal interpretations would lead to absurd and unintended results. To permit the continuation of custodial interrogation after a momentary cessation would clearly frustrate the purposes of Miranda by allowing repeated rounds of questioning to undermine the will of the person being questioned. At the other extreme, a blanket prohibition against the taking of voluntary statements or a permanent immunity from further interrogation, regardless of the circumstances, would transform the Miranda safeguards into wholly irrational obstacles to legitimate police investigative activity, and deprive suspects of an opportunity to make informed and intelligent assessments of their interests. Clearly, therefore, neither this passage nor any other passage in the Miranda opinion can sensibly be read to create a per se proscription of indefinite duration upon any further questioning by any *103police officer on any subject, once the person in custody has indicated a desire to remain silent.9

A reasonable and faithful interpretation of the Miranda opinion must rest on the intention of the Court in that case to adopt “fully effective means ... to notify the person of his right of silence and to assure that the exercise of the right will be scrupulously honored . . . 384 U. S., at 479. The critical safeguard identified in the passage at issue is a person’s “right to cut off questioning.” Id., at 474. Through the exercise of his option to terminate questioning he can control the time at *104which questioning occurs, the subjects discussed, and the duration of the interrogation. The requirement that law enforcement authorities must respect a person’s exercise of that option counteracts the coercive pressures of the custodial setting. We therefore conclude that the admissibility of statements obtained after the person in custody has decided to remain silent depends under Miranda on whether his “right to cut off questioning” was “scrupulously honored.” 10

A review of the circumstances leading to Mosley’s confession reveals that his “right to cut off questioning” was fully respected in this case. Before his initial interrogation, Mosley was carefully advised that he was under no obligation to answer any questions and could remain silent if he wished. He orally acknowledged that he understood the Miranda warnings and then signed a printed notification-of-rights form. When Mosley stated that he did not want to discuss the robberies, Detective Cowie immediately ceased the interrogation and did not try either to resume the questioning or in any way to persuade Mosley to reconsider his position. After an interval of more than two hours, Mosley was questioned by another police officer at another location about an unrelated holdup murder. He was given full and complete Miranda warnings at the outset of the second interrogation. He was thus reminded again that he could remain silent and could consult with a lawyer, *105and was carefully given a full and fair opportunity to exercise these options. The subsequent questioning did not undercut Mosley’s previous decision not to answer Detective Cowie’s inquiries. Detective Hill did not resume the interrogation about the White Tower Restaurant robbery or inquire about the Blue Goose Bar robbery, but instead focused exclusively on the Leroy Williams homicide, a crime different in nature and in time and place of occurrence from the robberies for which Mosley had been arrested and interrogated by Detective Cowie. Although it is not clear from the record how much Detective Hill knew about the earlier interrogation, his questioning of Mosley about an unrelated homicide was quite consistent with a reasonable interpretation of Mosley’s earlier refusal to answer any questions about the robberies.11

This is not a case, therefore, where the police failed to honor a decision of a person in custody to cut off questioning, either by refusing to discontinue the interrogation upon request or by persisting in repeated efforts to *106wear down his resistance and make him change his mind. In contrast to such practices, the police here immediately ceased the interrogation, resumed questioning only after the passage of a significant period of time and the provision of a fresh set of warnings, and restricted the second interrogation to a crime that had not been a subject of the earlier interrogation.

The Michigan Court of Appeals viewed this case as factually similar to Westover v. United States, 384 U. S. 436, a companion case to Miranda. But the controlling facts of the two cases are strikingly different.

In Westover, the petitioner was arrested by the Kansas City police at 9:45 p. m. and taken to the police station. Without giving any advisory warnings of any kind to Westover, the police questioned him that night and throughout the next morning about various local robberies. At noon, three FBI agents took over, gave advisory warnings to Westover, and proceeded to question him about two California bank robberies. After two hours of questioning, the petitioner confessed to the California crimes. The Court held that the confession obtained by the FBI was inadmissible because the interrogation leading to the petitioner’s statement followed on the heels of prolonged questioning that was commenced and continued by the Kansas City police without preliminary warnings to Westover of any kind. The Court found that “the federal authorities were the beneficiaries of the pressure applied by the local in-custody interrogation” and that the belated warnings given by the federal officers were “not sufficient to protect” West-over because from his point of view “the warnings came at the end of the interrogation process.” Id., at 497, 496.

Here, by contrast, the police gave full “Miranda warnings” to Mosley at the very outset of each interrogation, subjected him to only a brief period of initial question*107ing, and suspended questioning entirely for a significant period before beginning the interrogation that led to his incriminating statement. The cardinal fact of West-over — the failure of the police officers to give any warnings whatever to the person in their custody before embarking on an intense and prolonged interrogation of him — was simply not present in this case. The Michigan Court of Appeals was mistaken, therefore, in believing that Detective Hill’s questioning of Mosley was “not permitted” by the Westover decision. 51 Mich. App., at 108, 214 N. W. 2d, at 566.

For these reasons, we conclude that the admission in evidence of Mosley’s incriminating statement did not violate the principles of Miranda v. Arizona. Accordingly, the judgment of the Michigan Court of Appeals is vacated, and the case is remanded to that court for further proceedings not inconsistent with this opinion.

It is so ordered.

Mr. Justice White,

concurring in the result.

I concur in the result and in much of the majority’s reasoning. However, it appears to me that, in an effort to make only a limited holding in this case, the majority has implied that some custodial confessions will be suppressed even though they follow an informed and voluntary waiver of the defendant’s rights. The majority seems to say that a statement obtained within some unspecified time after an assertion by an individual of his “right to silence” is always inadmissible, even if it was the result of an informed and voluntary decision — following, for example, a disclosure to such an individual of a piece of information bearing on his waiver decision which the police had failed to give him prior to his assertion of the privilege but which they gave him immediately thereafter. Indeed, ante, at 102, the majority char*108acterizes as “absurd” any contrary rule. I disagree. I do not think the majority’s conclusion is compelled by Miranda v. Arizona, 384 U. S. 436 (1966), and I suspect that in the final analysis the majority will adopt voluntariness as the standard by which to judge the waiver of the right to silence by a properly informed defendant. I think the Court should say so now.

Miranda holds that custody creates an inherent compulsion on an individual to incriminate himself in response to questions, and that statements obtained under such circumstances are therefore obtained in violation of the Fifth Amendment privilege against compelled testimonial self-incrimination unless the privilege is “knowingly and intelligently waived.” Id., at 471, 475. It also holds that an individual will not be deemed to have made a knowing and intelligent waiver of his “right to silence” unless the authorities have first informed him, inter alia, of that right — “the threshold requirement for an intelligent decision as to its exercise.” Id., at 468. I am no more convinced that Miranda was required by the United States Constitution than I was when it was decided. However, there is at least some support in the law both before and after Miranda for the proposition that some rights will never be deemed waived unless the defendant is first expressly advised of their existence. E. g., Carnley v. Cochran, 369 U. S. 506 (1962); Boykin v. Alabama, 395 U. S. 238 (1969); Fed. Rules Crim. Proc. 11, 32 (a)(2). There is little support in the law or in common sense for the proposition that an informed waiver of a right may be ineffective even where voluntarily made. Indeed, the law is exactly to the contrary, e. g., Tollett v. Henderson, 411 U. S. 258 (1973); Brady v. United States, 397 U. S. 742 (1970); McMann v. Richardson, 397 U. S. 759 (1970); Parker v. North Carolina, 397 U. S. 790 (1970). Unless an individual is *109incompetent, we have in the past rejected any paternalistic rule protecting a defendant from his intelligent and voluntary decisions about his own criminal case. Faretta v. California, 422 U. S. 806 (1975). To do so would be to "imprison a man in his privileges,” 1 Adams v. United States ex rel. McCann, 317 U. S. 269, 280 (1942), and to disregard “ 'that respect for the individual which is the lifeblood of the law,’ ” Faretta v. California, supra, at 834. I am very reluctant to conclude that Miranda stands for such a proposition.

The language of Miranda no more compels such a result than does its basic rationale. As the majority points out, the statement in Miranda, 384 U. S., at 474, requiring interrogation to cease after an assertion of the “right to silence” tells us nothing because it does not indicate how soon this interrogation may resume. The Court showed in the very next paragraph, moreover, that when it wanted to create a per se rule against further interrogation after assertion of a right, it knew how to do so. The Court there said “[i]f the individual states that he *110wants an attorney, the interrogation must cease until an attorney is present.” Ibid. 2 However, when the individual indicates that he will decide unaided by counsel whether or not to assert his “right to silence” the situation is different. In such a situation, the Court in Miranda simply said: “If the interrogation continues without the presence of an attorney and a statement is taken, a heavy burden rests on the government to demonstrate that the defendant knowingly and intelligently waived his privilege against self-incrimination and his right to retained or appointed counsel.” Id., at 475. Apparently, although placing a heavy burden on the government, Miranda intended waiver of the “right to silence” to be tested by the normal standards. In any event, insofar as the Miranda decision might be read to require interrogation to cease for some magical and unspecified period of time following an assertion of the “right to silence,” and to reject voluntariness as the standard by which to judge informed waivers of that right, it should be disapproved as inconsistent with otherwise uniformly applied legal principles.

In justifying the implication that questioning must inevitably cease for some unspecified period of time following an exercise of the “right to silence,” the ma*111jority says only that such a requirement would be necessary to avoid •'‘undermining” “the will of the person being questioned.” Yet surely a waiver of the “right to silence” obtained by “undermining the will” of the person being questioned would be considered an involuntary waiver. Thus, in order to achieve the majority’s only stated purpose, it is sufficient to exclude all confessions which are the result of involuntary waivers. To exclude any others is to deprive the factfinding process of highly probative information for no reason at all. The “repeated rounds” of questioning following an assertion of the privilege, which the majority is worried about, would, of course, count heavily against the State in any determination of voluntariness — particularly if no reason (such as new facts communicated to the accused or a new incident being inquired about) appeared for repeated questioning. There is no reason, however, to rob the accused of the choice to answer questions voluntarily for some unspecified period of time following his own previous contrary decision. The Court should now so state.

Mr. Justice Brennan,

with whom Mr. Justice Marshall joins, dissenting.

The Court focuses on the correct passage from Miranda v. Arizona, 384 U. S. 436, 473-474 (1966) (footnote omitted):

“Once warnings have been given, the subsequent procedure is clear. If the individual indicates in any manner, at any time prior to or during questioning, that he wishes to remain silent, the interrogation must cease. At this point he has shown that he intends to exercise his Fifth Amendment privilege; any statement taken after the person invokes his privilege cannot be other than the product of compulsion, subtle or otherwise. Without the right to *112cut off questioning, the setting of in-custody interrogation operates on the individual to overcome free choice in producing a statement after the privilege has been once invoked.”

But the process of eroding Miranda rights, begun with Harris v. New York, 401 U. S. 222 (1971), continues with today’s holding that police may renew the questioning of a suspect who has once exercised his right to remain silent, provided the suspect’s right to cut off questioning has been “scrupulously honored.” Today’s distortion of Miranda’s constitutional principles can be viewed only as yet another step in the erosion and, I suppose, ultimate overruling of Miranda’s enforcement of the privilege against self-incrimination.

The Miranda guidelines were necessitated by the inherently coercive nature of in-custody questioning. As in Escobedo v. Illinois, 378 U. S. 478 (1964), “we sought a protective device to dispel the compelling atmosphere of the interrogation.” 384 U. S., at 465. We “concluded that without proper safeguards the process of in-custody interrogation of persons suspected or accused of crime contains inherently compelling pressures which work to undermine the individual’s will to resist and to compel him to speak where he would not otherwise do so freely.” Id., at 467.1 To assure safeguards that promised to dispel the “inherently compelling pressures” of in-custody interrogation, a prophylactic rule was fashioned to supplement the traditional determination of voluntariness on the facts of each case. Miranda held that any confession obtained when not preceded by the required warn*113ings or an adequate substitute safeguard was per se inadmissible in evidence. Id., at 468-469, 479. Satisfaction of this prophylactic rule, therefore, was necessary, though not sufficient, for the admission of a confession. Certiorari was expressly granted in Miranda “to give concrete constitutional guidelines for law enforcement agencies and courts to follow,” id., at 441-442, that is, clear, objective standards that might be applied to avoid the vagaries of the traditional voluntariness test.

The task that confronts the Court in this case is to satisfy the Miranda approach by establishing “concrete constitutional guidelines” governing the resumption of questioning a suspect who, while in custody, has once clearly and unequivocally “indicate [d] . . . that he wishes to remain silent . . . .” As the Court today continues to recognize, under Miranda, the cost of assuring voluntariness by procedural tests, independent of any actual inquiry into voluntariness, is that some voluntary statements will be excluded. Ante, at 99-100. Thus the consideration in the task confronting the Court is not whether voluntary statements will be excluded, but whether the procedures approved will be sufficient to assure with reasonable certainty that a confession is not obtained under the influence of the compulsion inherent, in interrogation and detention. The procedures approved by the Court today fail to provide that assurance.

We observed in Miranda: “Whatever the testimony of the authorities as to waiver of rights by an accused, the fact of lengthy interrogation or incommunicado incarceration before a statement is made is strong evidence that the accused did not validly waive his rights. In these circumstances the fact that the individual eventually made a statement is consistent with the conclusion that the compelling influence of the interrogation finally forced him to do so. It is inconsistent with any notion *114of a voluntary relinquishment of the privilege.” 384 U. S., at 476. And, as that portion of Miranda which the majority finds controlling observed, “the setting of in-custody interrogation operates on the individual to overcome free choice in producing a statement after the privilege has been once invoked.” Id., at 474. Thus, as to statements which are the product of renewed questioning, Miranda established a virtually irrebuttable presumption of compulsion, see id., at 474 n. 44, and that presumption stands strongest where, as in this case, a suspect, having initially determined to remain silent, is subsequently brought to confess his crime. Only by adequate procedural safeguards could the presumption be rebutted.

In formulating its procedural safeguard, the Court skirts the problem of compulsion and thereby fails to join issue with the dictates of Miranda. The language which the Court finds controlling in this ease teaches that renewed questioning itself is part of the process which invariably operates to overcome the will of a suspect. That teaching is embodied in the form of a proscription on any further questioning once the suspect has exercised his right to remain silent. Today’s decision uncritically abandons that teaching. The Court assumes, contrary to the controlling language, that “scrupulously honoring” an initial exercise of the right to remain silent preserves the efficaciousness of initial and future warnings despite the fact that the suspect has once been subjected to interrogation and then has been detained for a lengthy period of time.

Observing that the suspect can control the circumstances of interrogation “[tjhrough the exercise of his option to terminate questioning,” the Court concludes “that the admissibility of statements obtained after the person in custody has decided to remain silent depends ... *115on whether his ‘right to cut off questioning’ was ‘scrupulously honored.’ ” Ante, at 103, 104. But scrupulously honoring exercises of the right to cut off questioning is only meaningful insofar as the suspect’s will to exercise that right remains wholly unfettered. The Court’s formulation thus assumes the very matter at issue here: whether renewed questioning following a lengthy period of detention acts to overbear the suspect’s will, irrespective of giving the Miranda warnings a second time (and scrupulously honoring them), thereby rendering inconsequential any failure to exercise the right to remain silent. For the Court it is enough conclusorily to assert that “[t]he subsequent questioning did not undercut Mosley’s previous decision not to answer Detective Cowie’s inquiries.” Ante, at 105. Under Miranda, however, Mosley’s failure to exercise the right upon renewed questioning is presumptively the consequence of an overbearing in which detention and that subsequent questioning played central roles.

I agree that Miranda is not to be read, on the one hand, to impose an absolute ban on resumption of questioning “at any time or place on any subject,” ante, at 102, or on the other hand, “to permit a resumption of interrogation after a momentary respite,” ibid. But this surely cannot justify adoption of a vague and ineffective procedural standard that falls somewhere between those absurd extremes, for Miranda in flat and unambiguous terms requires that questioning “cease” when a suspect exercises the right to remain silent. Miranda’s terms, however, are not so uncompromising as to preclude the fashioning of guidelines to govern this case. Those guidelines must, of course, necessarily be sensitive to the reality that “[a]s a practical matter, the compulsion to speak in the isolated setting of the police station may well be greater than in courts or other official investiga*116tions, where there are often impartial observers to guard against intimidation or trickery.” 384 U. S., at 461 (footnote omitted).

The fashioning of guidelines for this case is an easy task. Adequate procedures are readily available. Michigan law requires that the suspect be arraigned before a judicial officer “without unnecessary delay,” 2 certainly not a burdensome requirement. Alternatively, a requirement that resumption of questioning should await appointment and arrival of counsel for the suspect would be an acceptable and readily satisfied precondition to resumption.3 Miranda expressly held that “[t]he presence of counsel . . . would be the adequate protective device necessary to make the process of police interrogation conform to the dictates of the privilege [against self-incrimination].” Id., at 466. The Court expediently bypasses this alternative in its search for circumstances where renewed questioning would be permissible.4

Indeed, language in Miranda suggests that the *117presence of counsel is the only appropriate alternative. In categorical language we held in Miranda: “If the individual indicates in any manner, at any time prior to or during questioning, that he wishes to remain silent, the interrogation must cease.” Id., at 473-474. We then immediately observed:

“If an individual indicates his desire to remain silent but has an attorney present, there may be some circumstances in which further questioning would be permissible. In the absence of evidence of overbearing, statements then made in the presence of counsel might be free of the compelling influence of the interrogation process and might fairly be construed as a waiver of the privilege for purposes of these statements.” Id., at 474 n. 44 (emphasis added).

This was the only circumstance in which we at all suggested that questioning could be resumed, and even then, further questioning was not permissible in all such circumstances, for compulsion was still the presumption not easily dissipated.5

*118These procedures would be wholly consistent with the Court’s rejection of a “per se proscription of indefinite duration,” ante, at 102, a rejection to which I fully subscribe. Today’s decision, however, virtually empties Miranda of principle, for plainly the decision encourages police asked to cease interrogation to continue the suspect’s detention until the police station’s coercive atmosphere does its work and the suspect responds to resumed questioning.6 Today’s rejection of that reality of life contrasts sharply with the Court’s acceptance only two years ago that “[i]n Miranda the Court found that the techniques of police questioning and the nature of custodial surroundings produce an inherently coercive situation.” Schneckloth v. Bustamonte, 412 U. S. 218, 247 (1973). I can only conclude that today’s decision signals rejection of Miranda’s basic premise.

My concern with the Court’s opinion does not end with its treatment of Miranda, but extends to its treatment of the facts in this case. The Court’s effort to have the Williams homicide appear as “an unrelated holdup murder,” ante, at 104, is patently unsuccessful. The anonymous tip received by Detective Cowie, conceded by the Court to be the sole basis for Mosley’s arrest, ante, at 97 n. 1, embraced both the robberies covered in Cowie’s in*119terrogation and the robbery-murder of Williams, ante, at 98 n. 2, about which Detective Hill questioned Mosley. Thus, when Mosley was apprehended, Cowie suspected him of being involved in the Williams robbery-murder in addition to the robberies about which he tried to examine Mosley. On another matter, the Court treats the second interrogation as being “at another location,” ante, at 104. Yet the fact is that it was merely a different floor of the same building, ante, at 97-98.7

I also find troubling the Court’s finding that Mosley never indicated that he did not want to discuss the robbery-murder, see ante, at 104-106. I cannot read Cowie’s testimony as the Court does. Cowie testified that Mosley *120declined to answer “ ‘[a'Jnything about the robberies/” ante, at 105 n. 11. That can be read only against the background of the anonymous tip that implicated Mosley in the Williams incident. Read in that light, it may reasonably be inferred that Cowie understood “[ajnything” to include the Williams episode, since the anonymous tip embraced that episode. More than this, the Court’s reading of Cowie’s testimony is not even faithful to the standard it articulates here today. “Anything about the robberies” may more than reasonably be interpreted as comprehending the Williams murder which occurred during a robbery. To interpret Mosley’s alleged statement to the contrary, therefore, hardly honors “scrupulously” the suspect’s rights.

In light of today’s erosion of Miranda standards as a matter of federal constitutional law, it is appropriate to observe that no State is precluded by the decision from adhering to higher standards under state law. Each State has power to impose higher standards governing police practices under state law than is required by the Federal Constitution. See Oregon v. Hass, 420 U. S. 714, 719 (1975);8 Lego v. Twomey, 404 U. S. 477, 489 (1972); Cooper v. California, 386 U. S. 58, 62 (1967). A decision particularly bearing upon the question of the adoption of Miranda as state law is Commonwealth v. Ware, 446 Pa. 52, 284 A. 2d 700 (1971). There the Pennsylvania Supreme Court adopted an aspect of Miranda as state law. This Court on March 20, *1211972, granted the Commonwealth’s petition for certiorari to review that decision. 405 U. S. 987. A month later, however, the error of the grant having been made apparent, the Court vacated the order of March 20, “it appearing that the judgment below rests upon an adequate state ground.” 406 U. S. 910. Understandably, state courts and legislatures are, as matters of state law, increasingly according protections once provided as federal rights but now increasingly depreciated by decisions of this Court. See, e. g., State v. Santiago, 53 Haw. 254, 492 P. 2d 657 (1971) (rejecting Harris v. New York, 401 U.S. 222 (1971)); People v. Beavers, 393 Mich. 554, 227 N. W. 2d 511 (1975), cert. denied, post, p. 878 (rejecting United States v. White, 401 U. S. 745 (1971)); State v. Johnson, 68 N. J. 349, 346 A. 2d 66 (1975) (rejecting Schneckloth v. Bustamonte, 412 U. S. 218 (1973)); Commonwealth v. Campana, 455 Pa. 622, 314 A. 2d 854, cert. denied, 417 U. S. 969 (1974) (adopting “same transaction or occurrence” view of Double Jeopardy Clause). I note that Michigan’s Constitution has its own counterpart to the privilege against self-incrimination. Mich. Const., Art. 1, § 17; see State v. Johnson, supra.

*454Justice O’Connor

delivered the opinion of the Court.

In Edwards v. Arizona, 451 U. S. 477 (1981), we held that law enforcement officers must immediately cease questioning a suspect who has clearly asserted his right to have counsel present during custodial interrogation. In this case we decide how law enforcement officers should respond when a suspect makes a reference to counsel that is insufficiently clear to invoke the Edwards prohibition on further questioning.

I

Pool brought trouble — not to River City, but to the Charleston Naval Base. Petitioner, a member of the United States Navy, spent the evening of October 2, 1988, shooting pool at a club on the base. Another sailor, Keith Shackleton, lost a game and a $30 wager to petitioner, but Shackleton refused to pay. After the club closed, Shackleton was beaten to death with a pool cue on a loading dock behind the commissary. The body was found early the next morning.

The investigation by the Naval Investigative Service (NIS) gradually focused on petitioner. Investigative agents determined that petitioner was at the club that evening, and that he was absent without authorization from his duty station the next morning. The agents also learned that only privately owned pool cues could be removed from the club premises, and that petitioner owned two cues — one of which had a bloodstain on it. The agents were told by various people that petitioner either had admitted committing the crime or had recounted details that clearly indicated his involvement in the killing.

On November 4, 1988, petitioner was interviewed at the NIS office. As required by military law, the agents advised petitioner that he was a suspect in the killing, that he was not required to make a statement, that any statement could be used against him at a trial by court-martial, and that he was entitled to speak with an attorney and have an attorney present during questioning. See Art. 31, Uniform Code of *455Military Justice (UCMJ), 10 U.S.C. §831; Mil. Rule Evid. 305; Manual for Courts-Martial A22-13 (1984). Petitioner waived his rights to remain silent and to counsel, both orally and in writing.

About an hour and a half into the interview, petitioner said, “Maybe I should talk to a lawyer.” App. 135. According to the uncontradicted testimony of one of the interviewing agents, the interview then proceeded as follows:

“[We m]ade it very clear that we’re not here to violate his rights, that if he wants a lawyer, then we will stop any kind of questioning with him, that we weren’t going to pursue the matter unless we have it clarified is he asking for a lawyer or is he just making a comment about a lawyer, and he said, [‘]No, I’m not asking for a lawyer,’ and then he continued on, and said, ‘No, I don’t want a lawyer.’” Id., at 136.

After a short break, the agents reminded petitioner of his rights to remain silent and to counsel. The interview then continued for another hour, until petitioner said, “I think I want a lawyer before I say anything else.” Id., at 137. At that point, questioning ceased.

At his general court-martial, petitioner moved to suppress statements made during the November 4 interview. The Military Judge denied the motion, holding that “the mention of a lawyer by [petitioner] during the course of the interrogation [was] not in the form of a request for counsel and . . . the agents properly determined that [petitioner] was not indicating a desire for or invoking his right to counsel.” Id., at 164. Petitioner was convicted on one specification of unpremeditated murder, in violation of Art. 118, UCMJ, 10 U. S. C. § 918. He was sentenced to confinement for life, a dishonorable discharge, forfeiture of all pay and allowances, and a reduction to the lowest pay grade. The convening authority approved the findings and sentence. The Navy-*456Marine Corps Court of Military Review affirmed. App. to Pet. for Cert. 12a-15a.

The United States Court of Military Appeals granted discretionary review and affirmed. 36 M. J. 337 (1993). The court recognized that the state and federal courts have developed three different approaches to a suspect’s ambiguous or equivocal request for counsel:

“Some jurisdictions have held that any mention of counsel, however ambiguous, is sufficient to require that all questioning cease. Others have attempted to define a threshold standard of clarity for invoking the right to counsel and have held that comments falling short of the threshold do not invoke the right to counsel. Some jurisdictions . . . have held that all interrogation about the offense must immediately cease whenever a suspect mentions counsel, but they allow interrogators to ask narrow questions designed to clarify the earlier statement and the [suspect’s] desires respecting counsel.” Id., at 341 (internal quotation marks omitted).

Applying the third approach, the court held that petitioner’s comment was ambiguous, and that the NIS agents properly clarified petitioner’s wishes with respect to counsel before continuing questioning him about the offense. Id., at 341-342.

Although we have twice previously noted the varying approaches the lower courts have adopted with respect to ambiguous or equivocal references to counsel during custodial interrogation, see Connecticut v. Barrett, 479 U. S. 523, 529-530, n. 3 (1987); Smith v. Illinois, 469 U. S. 91, 96, n. 3 (1984) (per curiam), we have not addressed the issue on the merits. We granted certiorari, 510 U. S. 942 (1993), to do so.

II

The Sixth Amendment right to counsel attaches only at the initiation of adversary criminal proceedings, see United *457 States v. Gouveia, 467 U. S. 180, 188 (1984), and before proceedings are initiated a suspect in a criminal investigation has no constitutional right to the assistance of counsel. Nevertheless, we held in Miranda v. Arizona, 384 U. S. 436, 469-473 (1966), that a suspect subject to custodial interrogation has the right to consult with an attorney and to have counsel present during questioning, and that the police must explain this right to him before questioning begins. The right to counsel established in Miranda was one of a “series of recommended ‘procedural safeguards’... [that] were not themselves rights protected by the Constitution but were instead measures to insure that the right against compulsory self-incrimination was protected.” Michigan v. Tucker, 417 U. S. 433, 443-444 (1974); see U. S. Const., Arndt. 5 (“No person . . . shall be compelled in any criminal case to be a witness against himself”).*

*458The right to counsel recognized in Miranda is sufficiently important to suspects in criminal investigations, we have held, that it “requires] the special protection of the knowing and intelligent waiver standard.” Edwards v. Arizona, 451 U. S., at 483. See Oregon v. Bradshaw, 462 U. S. 1039, 1046-1047 (1983) (plurality opinion); id., at 1051 (Powell, J., concurring in judgment). If the suspect effectively waives his right to counsel after receiving the Miranda warnings, law enforcement officers are free to question him. North Carolina v. Butler, 441 U. S. 369, 372-376 (1979). But if a suspect requests counsel at any time during the interview, he is not subject to further questioning until a lawyer has been made available or the suspect himself reinitiates conversation. Edwards v. Arizona, supra, at 484-485. This “second layer of prophylaxis for the Miranda right to counsel,” McNeil v. Wisconsin, 501 U. S. 171, 176 (1991), is “designed to prevent police from badgering a defendant into waiving his previously asserted Miranda rights,” Michigan v. Harvey, 494 U. S. 344, 350 (1990). To that end, we have held that a suspect who has invoked the right to counsel cannot be questioned regarding any offense unless an attorney is actually present. Minnick v. Mississippi, 498 U. S. 146 (1990); Arizona v. Roberson, 486 U. S. 675 (1988). “It remains clear, however, that this prohibition on further questioning — like other aspects of Miranda — is not itself required by the Fifth Amendment’s prohibition on coerced confessions, but is instead justified only by reference to its prophylactic purpose.” Connecticut v. Barrett, supra, at 528.

The applicability of the “ ‘rigid’ prophylactic rule” of Edwards requires courts to “determine whether the accused actually invoked his right to counsel.” Smith v. Illinois, supra, at 95 (emphasis added), quoting Fare v. Michael C., 442 U. S. 707, 719 (1979). To avoid difficulties of proof and to *459provide guidance to officers conducting interrogations, this is an objective inquiry. See Connecticut v. Barrett, supra, at 529. Invocation of the Miranda right to counsel “requires, at a minimum, some statement that can reasonably be construed to be an expression of a desire for the assistance of an attorney.” McNeil v. Wisconsin, 501 U. S., at 178. But if a suspect makes a reference to an attorney that is ambiguous or equivocal in that a reasonable officer in light of the circumstances would have understood only that the suspect might be invoking the right to counsel, our precedents do not require the cessation of questioning. See ibid. (“[T]he likelihood that a suspect would wish counsel to be present is not the test for applicability of Edwards”); Edwards v. Arizona, supra, at 485 (impermissible for authorities “to re-interrogate an accused in custody if he has clearly asserted his right to counsel”) (emphasis added).

Rather, the suspect must unambiguously request counsel. As we have observed, “a statement either is such an assertion of the right to counsel or it is not.” Smith v. Illinois, 469 U. S., at 97-98 (brackets and internal quotation marks omitted). Although a suspect need not “speak with the discrimination of an Oxford don,” post, at 476 (Souter, J., concurring in judgment), he must articulate his desire to have counsel present sufficiently clearly that a reasonable police officer in the circumstances would understand the statement to be a request for an attorney. If the statement fails to meet the requisite level of clarity, Edwards does not require that the officers stop questioning the suspect. See Moran v. Burbine, 475 U. S. 412, 433, n. 4 (1986) (“[T]he interrogation must cease until an attorney is present only [i]f the individual states that he wants an attorney”) (citations and internal quotation marks omitted).

We decline petitioner’s invitation to extend Edwards and require law enforcement officers to cease questioning immediately upon the making of an ambiguous or equivocal reference to an attorney. See Arizona v. Roberson, supra, at 688 *460(Kennedy, J., dissenting) (“[T]he rule of Edwards is our rule, not a constitutional command; and it is our obligation to justify its expansion”). The rationale underlying Edwards is that the police must respect a suspect’s wishes regarding his right to have an attorney present during custodial interrogation. But when the officers conducting the questioning reasonably do not know whether or not the suspect wants a lawyer, a rule requiring the immediate cessation of questioning “would transform the Miranda safeguards into wholly irrational obstacles to legitimate police investigative activity,” Michigan v. Mosley, 423 U. S. 96, 102 (1975), because it would needlessly prevent the police from questioning a suspect in the absence of counsel even if the suspect did not wish to have a lawyer present. Nothing in Edwards requires the provision of counsel to a suspect who consents to answer questions without the assistance of a lawyer. In Miranda itself, we expressly rejected the suggestion “that each police station must have a ‘station house lawyer’ present at all times to advise prisoners,” 384 U. S., at 474, and held instead that a suspect must be told of his right to have an attorney present and that he may not be questioned after invoking his right to counsel. We also noted that if a suspect is “indecisive in his request for counsel,” the officers need not always cease questioning. See id., at 485.

We recognize that requiring a clear assertion of the right to counsel might disadvantage some suspects who — because of fear, intimidation, lack of linguistic skills, or a variety of other reasons — will not clearly articulate their right to counsel although they actually want to have a lawyer present. But the primary protection afforded suspects subject to custodial interrogation is the Miranda warnings themselves. “[F]ull comprehension of the rights to remain silent and request an attorney [is] sufficient to dispel whatever coercion is inherent in the interrogation process.” Moran v. Burbine, supra, at 427. A suspect who knowingly and voluntarily waives his right to counsel after having that right explained *461to him has indicated his willingness to deal with the police unassisted. Although Edwards provides an additional protection — if a suspect subsequently requests an attorney, questioning must cease — it is one that must be affirmatively invoked by the suspect.

In considering how a suspect must invoke the right to counsel, we must consider the other side of the Miranda equation: the need for effective law enforcement. Although the courts ensure compliance with the Miranda requirements through the exclusionary rule, it is police officers who must actually decide whether or not they can question a suspect. The Edwards rule — questioning must cease if the suspect asks for a lawyer — provides a bright line that can be applied by officers in the real world of investigation and interrogation without unduly hampering the gathering of information. But if we were to require questioning to cease if a suspect makes a statement that might be a request for an attorney, this clarity and ease of application would be lost. Police officers would be forced to make difficult judgment calls about whether the suspect in fact wants a lawyer even though he has not said so, with the threat of suppression if they guess wrong. We therefore hold that, after a knowing and voluntary waiver of the Miranda rights, law enforcement officers may continue questioning until and unless the suspect clearly requests an attorney.

Of course, when a suspect makes an ambiguous or equivocal statement it will often be good police practice for the interviewing officers to clarify whether or not he actually wants an attorney. That was the procedure followed by the NIS agents in this case. Clarifying questions help protect the rights of the suspect by ensuring that he gets an attorney if he wants one, and will minimize the chance of a confession being suppressed due to subsequent judicial second-guessing as to the meaning of the suspect’s statement regarding counsel. But we decline to adopt a rule requiring officers to ask clarifying questions. If the suspect’s state*462ment is not an unambiguous or unequivocal request for counsel, the officers have no obligation to stop questioning him.

To recapitulate: We held in Miranda that a suspect is entitled to the assistance of counsel during custodial interrogation even though the Constitution does not provide for such assistance. We held in Edwards that if the suspect invokes the right to counsel at any time, the police must immediately cease questioning him until an attorney is present. But we are unwilling to create a third layer of prophylaxis to prevent police questioning when the suspect might want a lawyer. Unless the suspect actually requests an attorney, questioning may continue.

The courts below found that petitioner’s remark to the NIS agents — “Maybe I should talk to a lawyer” — was not a request for counsel, and we see no reason to disturb that conclusion. The NIS agents therefore were not required to stop questioning petitioner, though it was entirely proper for them to clarify whether petitioner in fact wanted a lawyer. Because there is no ground for suppression of petitioner’s statements, the judgment of the Court of Military Appeals is

Affirmed.

Justice Scalia,

concurring.

Section 3501 of Title 18 of the United States Code is “the statute governing the admissibility of confessions in federal prosecutions.” United States v. Alvarez-Sanchez, 511 U. S. 350, 351 (1994). That provision declares that “a confession . . . shall be admissible in evidence if it is voluntarily given," and that the issue of voluntariness shall be determined on the basis of “all the circumstances surrounding the giving of the confession, including whether or not [the] defendant was advised or knew that he was not required to make any statement...[;]... whether or not [the] defendant had been advised prior to questioning of his right to the assistance of counsel; and . . . whether or not [the] defendant was without the assistance of counsel when questioned____” *463§§ 3501(a), (b) (emphases added). It continues (lest the import be doubtful): “The presence or absence of any of the above-mentioned factors . . . need not be conclusive on the issue of voluntariness of the confession.” § 3501(b). Legal analysis of the admissibility of a confession without reference to these provisions is equivalent to legal analysis of the admissibility of hearsay without consulting the Rules of Evidence; it is an unreal exercise. Yet as the Court observes, see ante, at 457-458, n., that is precisely what the United States has undertaken in this case. It did not raise § 3501(a) below and asserted that it is “not at issue” here, Brief for United States 18, n. 13.*

This is not the first case in which the United States has declined to invoke § 3501 before us — nor even the first case in which that failure has been called to its attention. See Tr. of Oral Arg. in United States v. Green, O. T. 1992, No. 91-1521, pp. 18-21. In fact, with limited exceptions the *464provision has been studiously avoided by every Administration, not only in this Court but in the lower courts, since its enactment more than 25 years ago. See Office of Legal Policy, U. S. Dept, of Justice, Report to Attorney General on Law of Pre-Trial Interrogation 72-73 (1986) (discussing “[t]he abortive implementation of §3501” after its passage in 1968).

I agree with the Court that it is proper, given the Government’s failure to raise the point, to render judgment without taking account of §3501. But the refusal to consider arguments not raised is a sound prudential practice, rather than a statutory or constitutional mandate, and there are times when prudence dictates the contrary. See United States Nat. Bank of Ore. v. Independent Ins. Agents of America, Inc., 508 U. S. 439, 445-448 (1993) (proper for Court of Appeals to consider whether an allegedly controlling statute had been repealed, despite parties’ failure, upon invitation, to assert the point). As far as I am concerned, such a time will have arrived when a case that comes within the terms of this statute is next presented to us.

For most of this century, voluntariness vel non was the touchstone of admissibility of confessions. See Miranda v. Arizona, 384 U. S. 436, 506-507 (1966) (Harlan, J., dissenting). Section 3501 of Title 18 seems to provide for that standard in federal criminal prosecutions today. I. say “seems” because I do not wish to prejudge any issue of law. I am entirely open to the argument that § 3501 does not mean what it appears to say; that it is inapplicable for some other reason; or even that it is unconstitutional. But I will no longer be open to the argument that this Court should continue to ignore the commands of §3501 simply because the Executive declines to insist that we observe them.

The Executive has the power (whether or not it has the right) effectively to nullify some provisions of law by the mere failure to prosecute — the exercise of so-called prosecutorial discretion. And it has the power (whether or not it *465has the right) to avoid application of §3501 by simply declining to introduce into evidence confessions admissible under its terms. But once a prosecution has been commenced and a confession introduced, the Executive assuredly has neither the power nor the right to determine what objections to admissibility of the confession are valid in law. Section § 3501 of Title 18 is a provision of law directed to the courts, reflecting the people’s assessment of the proper balance to be struck between concern for persons interrogated in custody and the needs of effective law enforcement. We shirk our duty if we systematically disregard that statutory command simply because the Justice Department systematically declines to remind us of it.

The United States’ repeated refusal to invoke §3501, combined with the courts’ traditional (albeit merely prudential) refusal to consider arguments not raised, has caused the federal judiciary to confront a host of “Miranda” issues that might be entirely irrelevant under federal law. See, e. g., in addition to the present case, United States v. Green, 507 U. S. 545 (1993) (dism’g cert. as moot); United States v. Griffin, 922 F. 2d 1343 (CA8 1990); United States v. Vazquez, 857 F. 2d 857 (CA1 1988); United States v. Scalf 725 F. 2d 1272 (CA10 1984). Worse still, it may have produced — during an era of intense national concern about the problem of runaway crime — the acquittal and the nonprosecution of many dangerous felons, enabling them to continue their depredations upon our citizens. There is no excuse for this. Perhaps (though I do not immediately see why) the Justice Department has good basis for believing that allowing prosecutions to be defeated on grounds that could be avoided by invocation of § 3501 is consistent with the Executive’s obligation to “take Care that the Laws be faithfully executed,” U. S. Const., Art. II, §3. That is not the point. The point is whether our continuing refusal to consider § 3501 is consistent with the Third Branch’s obligation to decide according to the law. I think it is not.

*466Justice Souter,

with whom Justice Blackmun, Justice Stevens, and Justice Ginsburg join, concurring in the judgment.

In the midst of his questioning by naval investigators, petitioner said “Maybe I should talk to a lawyer.” The investigators promptly stopped questioning Davis about the killing of Keith Shackleton and instead undertook to determine whether he meant to invoke his right' to counsel, see Miranda v. Arizona, 384 U. S. 436 (1966). According to testimony accepted by the courts below, Davis answered the investigators’ questions on that point by saying, “I’m not asking for a lawyer,” and “No, I don’t want to talk to a lawyer.” Only then did the interrogation resume (stopping for good when petitioner said, “I think I want a lawyer before I say anything else”).

I agree with the majority that the Constitution does not forbid law enforcement officers to pose questions (like those directed at Davis) aimed solely at clarifying whether a suspect’s ambiguous reference to counsel was meant to assert his Fifth Amendment right. Accordingly I concur in the judgment affirming Davis’s conviction, resting partly on evidence of statements given after agents ascertained that he did not wish to deal with them through counsel. I cannot, however, join in my colleagues’ further conclusion that if the investigators here had been so inclined, they were at liberty to disregard Davis’s reference to a lawyer entirely, in accordance with a general rule that interrogators have no legal obligation to discover what a custodial subject meant by an ambiguous statement that could reasonably be understood to express a desire to consult a lawyer.

Our own precedent, the reasonable judgments of the majority of the many courts already to have addressed the issue before us,1 and the advocacy of a considerable body of law *467enforcement officials2 are to the contrary. All argue against the Court’s approach today, which draws a sharp line between interrogated suspects who “clearly” assert their right to counsel, ante, at 461, and those who say something that may, but may not, express a desire for counsel’s presence, the former suspects being assured that questioning will not resume without counsel present, see Miranda, supra, at 474, Edwards v. Arizona, 451 U. S. 477, 484-485 (1981); Minnick v. Mississippi, 498 U. S. 146 (1990), the latter being left to fend for themselves. The concerns of fairness and practicality that have long anchored our Miranda case law point to a different response: when law enforcement officials “reasonably do not know whether or not the suspect wants a lawyer,” ante, at 460, they should stop their interrogation and ask him to make his choice clear.

I

A

While the question we address today is an open one,3 its answer requires coherence with nearly three decades of case *468law addressing the relationship between police and criminal suspects in custodial interrogation. Throughout that period, two precepts have commanded broad assent: that the *469 Miranda safeguards exist “ ‘to assure that the individual’s right to choose between speech and silence remains unfettered throughout the interrogation process,’ ” see Connecticut v. Barrett, 479 U. S. 523, 528 (1987) (quoting Miranda, 384 U. S., at 469, and supplying emphasis), and that the justification for Miranda rules, intended to operate in the real world, “must be consistent with ... practical realities,” Arizona v. Roberson, 486 U. S. 675, 688 (1988) (Kennedy, J., dissenting). A rule barring government agents from further interrogation until they determine whether a suspect’s ambiguous statement was meant as a request for counsel fulfills both ambitions. It assures that a suspect’s choice whether or not to deal with police through counsel will be “scrupulously honored,” Miranda, supra, at 479; cf. Michigan v. Mosley, 423 U. S. 96, 110, n. 2 (1975) (White, J., concurring in result), and it faces both the real-world reasons why misunderstandings arise between suspect and interrogator and the real-world limitations on the capacity of police and trial courts to apply fine distinctions and intricate rules.

B

Tested against the same two principles, the approach the Court adopts does not fare so well. First, as the majority expressly acknowledges, see ante, at 460, criminal suspects who may (in Miranda’s words) be “thrust into an unfamiliar atmosphere and run through menacing police interrogation procedures,” 384 U. S., at 457, would seem an odd group to single out for the Court’s demand of heightened linguistic care. A substantial percentage of them lack anything like a confident command of the English language, see, e. g., United States v. De la Jara, 973 F. 2d 746, 750 (CA9 1992); many are “woefully ignorant,” Miranda, supra, at 468; cf. Davis v. North Carolina, 384 U. S. 737, 742 (1966); and many more will be sufficiently intimidated by the interrogation process or overwhelmed by the uncertainty of their predicament that *470the ability to speak assertively will abandon them.4 Indeed, the awareness of just these realities has, in the past, dissuaded the Court from placing any burden of clarity upon individuals in custody, but has led it instead to require that requests for counsel be “give[n] a broad, rather than a narrow, interpretation,” see Michigan v. Jackson, 475 U. S. 625, 633 (1986); Barrett, supra, at 529, and that courts “indulge every reasonable presumption,” Johnson v. Zerbst, 304 U. S. 458, 464 (1938) (internal quotation marks omitted), that a suspect has not waived his right to counsel under Miranda, see, e. g., Oregon v. Bradshaw, 462 U. S. 1039, 1051 (1983) (Powell, J., concurring) (“We are unanimous in agreeing . . . that the [Miranda] right to counsel is a prime example of those rights requiring the special protection of the knowing and intelligent waiver standard”) (internal quotation marks and brackets omitted); cf. Minnick, 498 U. S., at 160 (Scalia, J., dissenting) (“[W]e have adhered to the principle that nothing less than the Zerbst standard” is appropriate for Miranda waivers).

Nor may the standard governing waivers as expressed in these statements be deflected away by drawing a distinction between initial waivers of Miranda rights and subsequent *471decisions to reinvoke them, on the theory that so long as the burden to demonstrate waiver rests on the government, it is only fair to make the suspect shoulder a burden of showing a clear subsequent assertion. Miranda itself discredited the legitimacy of any such distinction. The opinion described the object of the warning as being to assure “a continuous opportunity to exercise [the right of silence],” 384 U. S., at 444; see also Moran v. Burbine, 475 U. S. 412, 458 (1986) (Stevens, J., dissenting); accord, id., at 423, n. 1. “[Continuous opportunity” suggests an unvarying one, governed by a common standard of effectiveness. The suggestion is confirmed by the very first statement that follows, that “there can be no questioning” if the suspect “indicates in any manner and at any stage of the process that he wishes to consult with an attorney,” Miranda, 384 U. S., at 444-445. “[A]t any stage” obviously includes the stage after initial waiver and the commencement of questioning, and “indicates in any manner” is a rule plainly in tension with the indication “with a vengeance,” see id., at 505 (Harlan, J., dissenting), that the Court would require for exercise of the “continuous” right at some point after initial waiver.

The Court defends as tolerable the certainty that some poorly expressed requests for counsel will be disregarded on the ground that Miranda warnings suffice to alleviate the inherent coercion of the custodial interrogation. Ante, at 460. But, “[a] once-stated warning, delivered by those who will conduct the interrogation, cannot itself suffice” to “assure that the . . . right to choose between silence and speech remains unfettered throughout the interrogation process,” 384 U. S., at 469. Nor does the Court’s defense reflect a sound reading of the case it relies on, Moran v. Burbine, supra:

“Beyond [the] duty to inform, Miranda requires that the police respect the [suspect’s] decision to exercise the rights outlined in the warnings. ‘If the individual indicates in any manner, at any time prior to or during ques*472tioning, that he wishes to remain silent, [or if he] states that he wants an attorney, the interrogation must cease.’” 475 U. S., at 420 (quoting Miranda, supra, at 473-474).

While Moran held that a subject’s knowing and voluntary waiver of the right to counsel is not undermined by the fact, that police prevented an unsummoned lawyer from making contact with him, it contains no suggestion that Miranda affords as ready a tolerance for police conduct frustrating the suspect’s subjectively held (if ambiguously expressed) desire for counsel. See 475 U. S., at 423 (contrasting Escobedo v. Illinois, 378 U. S. 478, 481 (1964), where “police incorrectly told the suspect that his lawyer ‘didn’t want to see him’ ”); see also Miranda, supra, at 468 (purpose of warnings is to “show the individual that his interrogators are prepared to recognize his privilege should he choose to exercise it”).

Indeed, it is easy, amidst the discussion of layers of protection, to lose sight of a real risk in the majority’s approach, going close to the core of what the Court has held that the Fifth Amendment provides. The experience of the timid or verbally inept suspect (whose existence the Court acknowledges) may not always closely follow that of the defendant in Edwards v. Arizona (whose purported waiver of his right to counsel, made after having invoked the right, was held ineffective, lest police be tempted to “badge[r]” others like him, see Michigan v. Harvey, 494 U. S. 344, 350 (1990)). Indeed, it may be more like that of the defendant in Escobedo v. Illinois, supra, whose sense of dilemma was heightened by his interrogators’ denial of his requests to talk to a lawyer. When a suspect understands his (expressed) wishes tó have been ignored (and by hypothesis, he has said something that an objective listener could “reasonably,” although not necessarily, take to be a request), in contravention of the “rights” just read to him by his interrogator, he may well *473see further objection as futile and confession (true or not) as the only way to end his interrogation.5

Nor is it enough to say that a “ ‘statement either is ... an assertion of the right to counsel or it is not.’ ” Ante, at 459 (quoting Smith v. Illinois, 469 U. S., at 97-98) (omitting brackets and internal quotation marks). In Smith, we neither denied the possibility that a reference to counsel could be ambiguous, see id., at 98; accord, id., at 101 (Rehnquist, J., dissenting), nor suggested that particular statements should be considered in isolation, id., at 98.6 While it might be fair to say that every statement is meant either to express a desire to deal with police through counsel or not, this fact does not dictate the rule that interrogators who hear a statement consistent with either possibility may presume the latter and forge ahead; on the contrary, clarification is the intuitively sensible course.

The other justifications offered for the “requisite level of clarity” rule, ante, at 459, are that, whatever its costs, it will further society’s strong interest in “effective law enforcement,” ante, at 461, and maintain the “ease of application,” *474 ibid., that has long been a concern of our Miranda jurisprudence. With respect to the first point, the margin of difference between the clarification approach advocated here and the one the Court adopts is defined by the class of cases in which a suspect, if asked, would make it plain that he meant to request counsel (at which point questioning would cease). While these lost confessions do extract a real price from society, it is one that Miranda itself determined should be borne. Cf. Brief for Americans for Effective Law Enforcement, Inc., et al. as Amici Curiae 5 (the clarification approach “preserves the interests of law enforcement and of the public welfare”); Escobedo, supra, at 490 (“No system worth preserving should have to fear that if an accused is permitted to consult with a lawyer, he will become aware of, and exercise, [his constitutional] rights”).

As for practical application, while every approach, including the majority’s, will involve some “difficult judgment calls,”7 the rule argued for here would relieve the officer of *475any responsibility for guessing “whether the suspect in fact wants a lawyer even though he hasn’t said so,” ante, at 461. To the contrary, it would assure that the “judgment call” will be made by the party most competent to resolve the ambiguity, who our case law has always assumed should make it: the individual suspect.

II

Although I am convinced that the Court has taken the wrong path, I am not persuaded by petitioner’s contention that even ambiguous statements require an end to all police questioning. I recognize that the approach petitioner urges on us can claim some support from our case law, most notably in the “indicates in any manner” language of Miranda, and I do not deny that the rule I endorse could be abused by “clarifying” questions that shade subtly into illicitly badgering a suspect who wants counsel, but see Thompson v. Wainwright, 601 F. 2d 768, 771-772 (CA5 1979); cf. State v. Walkowiak, 183 Wis. 2d 478, 515 N. W. 2d 863 (1994) (Abrahamson, J., concurring) (suggesting means properly to focus clarification enquiry). But petitioner’s proposal is not entirely in harmony with all the major themes of Miranda case law, its virtues and demerits being the reverse images of those that mark the Court’s rule. While it is plainly wrong, for example, to continue interrogation when the suspect wants it to stop (and so indicates), the strong bias in favor of individual choice may also be disserved by stopping questioning when a suspect wants it to continue (but where his statement might be understood otherwise), see Michigan v. *476 Mosley, 423 U. S. 96,109 (1975) (White, J., concurring in result) (“[W]e have... rejected [the] paternalistic rule protecting a defendant from his intelligent and voluntary decisions about his own criminal case”). The costs to society of losing confessions would, moreover, be especially hard to bear where the suspect, if asked for his choice, would have chosen to continue. One need not sign the majority’s opinion here to agree that resort to the rule petitioner argues for should be had only if experience shows that less drastic means of safeguarding suspects’ constitutional rights are not up to the job, see generally United States v. Leon, 468 U. S. 897, 927-928 (1984) (Blackmun, J., concurring) (exclusionary rule exception must be “tested in the real world of state and federal law enforcement, and this Court will attend to the results”).

* * *

Our cases are best respected by a rule that when a suspect under custodial interrogation makes an ambiguous statement that might reasonably be understood as expressing a wish that a lawyer be summoned (and questioning cease), interrogators’ questions should be confined to verifying whether the individual meant to ask for a lawyer. While there is reason to expect that trial courts will apply today’s ruling sensibly (without requiring criminal suspects to speak with the discrimination of an Oxford don) and that interrogators will continue to follow what the Court rightly calls “good police practice” (compelled up to now by a substantial body of state and Circuit law), I believe that the case law under Miranda does not allow them to do otherwise.

Justice Scalia

delivered the opinion of the Court.

We consider whether a break in custody ends the presumption of involuntariness established in Edwards v. Arizona, 451 U. S. 477 (1981).

I

In August 2003, a social worker assigned to the Child Advocacy Center in the Criminal Investigation Division of the Hagerstown Police Department referred to the department allegations that respondent Michael Shatzer, Sr., had sexually abused his 3-year-old son. At that time, Shatzer was *101incarcerated at the Maryland Correctional InstitutionHagerstown, serving a sentence for an unrelated child-sexual-abuse offense. Detective Shane Blankenship was assigned to the investigation and interviewed Shatzer at the correctional institution on August 7, 2003. Before asking any questions, Blankenship reviewed Shatzer’s Miranda rights with him, and obtained a written waiver of those rights. When Blankenship explained that he was there to question Shatzer about sexually abusing his son, Shatzer expressed confusion — he had thought Blankenship was an attorney there to discuss the prior crime for which he was incarcerated. Blankenship clarified the purpose of his visit, and Shatzer declined to speak without an attorney. Accordingly, Blankenship ended the interview, and Shatzer was released back into the general prison population. Shortly thereafter, Blankenship closed the investigation.

Two years and six months later, the same social worker referred more specific allegations to the department about the same incident involving Shatzer. Detective Paul Hoover, from the same division, was assigned to the investigation. He and the social worker interviewed the victim, then eight years old, who described the incident in more detail. With this new information in hand, on March 2, 2006, they went to the Roxbury Correctional Institute, to which Shatzer had since been transferred, and interviewed Shatzer in a maintenance room outfitted with a desk and three chairs. Hoover explained that he wanted to ask Shatzer about the alleged incident involving Shatzer’s son. Shatzer was surprised because he thought that the investigation had been closed, but Hoover explained they had opened a new file. Hoover then read Shatzer his Miranda rights and obtained a written waiver on a standard department form.

Hoover interrogated Shatzer about the incident for approximately 30 minutes. Shatzer denied ordering his son to perform fellatio on him, but admitted to masturbating in *102front of his son from a distance of less than three feet. Before the interview ended, Shatzer agreed to Hoover’s request that he submit to a polygraph examination. At no point during the interrogation did Shatzer request to speak with an attorney or refer to his prior refusal to answer questions without one.

Five days later, on March 7, 2006, Hoover and another detective met with Shatzer at the correctional facility to administer the polygraph examination. After reading Shatzer his Miranda rights and obtaining a written waiver, the other detective administered the test and concluded that Shatzer had failed. When the detectives then questioned Shatzer, he became upset, started to cry, and incriminated himself by saying, “ ¶ didn’t force him. I didn’t force him.’ ” 405 Md. 585, 590, 954 A. 2d 1118,1121 (2008). After making this inculpatory statement, Shatzer requested an attorney, and Hoover promptly ended the interrogation.

The State’s Attorney for Washington County charged Shatzer with second-degree sexual offense, sexual child abuse, second-degree assault, and contributing to conditions rendering a child in need of assistance. Shatzer moved to suppress his March 2006 statements pursuant to Edwards. The trial court held a suppression hearing and later denied Shatzer’s motion. The Edwards protections did not apply, it reasoned, because Shatzer had experienced a break in custody for Miranda purposes between the 2003 and 2006 interrogations. No. 21-K-06-37799 (Cir. Ct. Washington Cty., Md., Sept. 14, 2006), App. 55. Shatzer pleaded not guilty, waived his right to a jury trial, and proceeded to a bench trial based on an agreed statement of facts. In accordance with the agreement, the State described the interview with the victim and Shatzer’s 2006 statements to the detectives. Based on the proffered testimony of the victim and the “admission of the defendant as to the act of masturbation,” the trial court found Shatzer guilty of sexual child abuse of his *103son.1 No. 21-K-06-37799 (Cir. Ct. Washington Cty., Md., Sept. 21, 2006), id., at 70, 79.

Over the dissent of two judges, the Court of Appeals of Maryland reversed and remanded. The court held that “the passage of time alone is insufficient to [end] the protections afforded by Edwards,” and that, assuming, arguendo, a break-in-custody exception to Edwards existed, Shatzer’s release back into the general prison population between interrogations did not constitute a break in custody. 405 Md., at 606-607,954 A. 2d, at 1131. We granted certiorari, 555 U. S. 1152 (2009).

II

The Fifth Amendment, which applies to the States by virtue of the Fourteenth Amendment, Malloy v. Hogan, 378 U. S. 1, 6 (1964), provides that “[n]o person ... shall be compelled in any criminal case to be a witness against himself.” U. S. Const., Arndt. 5. In Miranda v. Arizona, 384 U. S. 436 (1966), the Court adopted a set of prophylactic measures to protect a suspect’s Fifth Amendment right from the “inherently compelling pressures” of custodial interrogation. Id., at 467. The Court observed that “incommunicado interrogation” in an “unfamiliar,” “police-dominated atmosphere,” id., at 456-457, involves psychological pressures “which work to undermine the individual’s will to resist and to compel him to speak where he would not otherwise do so freely,” id., at 467. Consequently, it reasoned, “[u]nless adequate protective devices are employed to dispel the compulsion inherent in custodial surroundings, no statement obtained from the defendant can truly be the product of his free choice.” Id., at 458.

To counteract the coercive pressure, Miranda announced that police officers must warn a suspect prior to questioning *104that he has a right to remain silent, and a right to the presence of an attorney. Id., at 444. After the warnings are given, if the suspect indicates that he wishes to remain silent, the interrogation must cease. Id., at 473-474. Similarly, if the suspect states that he wants an attorney, the interrogation must cease until an attorney is present. Id., at 474. Critically, however, a suspect can waive these rights. Id., at 475. To establish a valid waiver, the State must show that the waiver was knowing, intelligent, and voluntary under the “high standard] of proof for the waiver of constitutional rights [set forth in] Johnson v. Zerbst, 304 U. S. 458 (1938).” Id., at 475.

In Edwards, the Court determined that Zerbsfs traditional standard for waiver was not sufficient to protect a suspect’s right to have counsel present at a subsequent interrogation if he had previously requested counsel; “additional safeguards” were necessary. 451 U. S., at 484. The Court therefore superimposed a “second layer of prophylaxis,” McNeil v. Wisconsin, 501 U. S. 171, 176 (1991). Edwards held:

“[W]hen an accused has invoked his right to have counsel present during custodial interrogation, a valid waiver of that right cannot be established by showing only that he responded to further police-initiated custodial interrogation even if he has been advised of his rights. . . . [He] is not subject to further interrogation by the authorities until counsel has been made available to him, unless the accused himself initiates further communication, exchanges, or conversations with the police.” 451 U. S., at 484-485.

The rationale of Edwards is that once a suspect indicates that “he is not capable of undergoing [custodial] questioning without advice of counsel,” “any subsequent waiver that has come at the authorities’ behest, and not at the suspect’s own instigation, is itself the product of the ‘inherently compelling *105pressures’ and not the purely voluntary choice of the suspect.” Arizona v. Roberson, 486 U. S. 675, 681 (1988). Under this rule, a voluntary Miranda waiver is sufficient at the time of an initial attempted interrogation to protect a suspect’s right to have counsel present, but it is not sufficient at the time of subsequent attempts if the suspect initially requested the presence of counsel. The implicit assumption, of course, is that the subsequent requests for interrogation pose a significantly greater risk of coercion. That increased risk results not only from the police’s persistence in trying to get the suspect to talk, but also from the continued pressure that begins when the individual is taken into custody as a suspect and sought to be interrogated — pressure likely to “increase as custody is prolonged,” Minnick v. Mississippi, 498 U. S. 146, 153 (1990). The Edwards presumption of involuntariness ensures that police will not take advantage of the mounting coercive pressures of “prolonged police custody,” Roberson, 486 U. S., at 686, by repeatedly attempting to question a suspect who previously requested counsel until the suspect is “badgered into submission,” id., at 690 (Kennedy, J., dissenting).

We have frequently emphasized that the Edwards rule is not a constitutional mandate, but judicially prescribed prophylaxis. See, e. g., Montejo v. Louisiana, 556 U. S. 778, 787 (2009); Michigan v. Harvey, 494 U. S. 344, 349 (1990); Solem v. Stumes, 465 U. S. 638, 644, n. 4 (1984). Because Edwards is “our rule, not a constitutional command,” “it is our obligation to justify its expansion.” Roberson, supra, at 688 (Kennedy, J., dissenting). Lower courts have uniformly held that a break in custody ends the Edwards presumption, see, e. g., People v. Storm, 28 Cal. 4th 1007, 1023-1024, and n. 6, 52 P. 3d 52, 61-62, and n. 6 (2002) (collecting state and federal cases), but we have previously addressed the issue only in dicta, see McNeil, supra, at 177 (Edwards applies “assuming there has been no break in custody”).

*106A judicially crafted rule is “justified only by reference to its prophylactic purpose,” Davis v. United States, 512 U. S. 452, 458 (1994) (internal quotation marks omitted), and applies only where its benefits outweigh its costs, Montejo, supra, at 793. We begin with the benefits. Edwards’ presumption of involuntariness has the incidental effect of “conserving] judicial resources which would otherwise be expended in making difficult determinations of voluntariness.” Minnick, supra, at 151. Its fundamental purpose, however, is to “[p]reserv[e] the integrity of an accused’s choice to communicate with police only through counsel,” Patterson v. Illinois, 487 U. S. 285, 291 (1988), by “preventing] police from badgering a defendant into waiving his previously asserted Miranda rights,” Harvey, supra, at 350. Thus, the benefits of the rule are measured by the number of coerced confessions it suppresses that otherwise would have been admitted. See Montejo, supra, at 793.

It is easy to believe that a suspect may be coerced or badgered into abandoning his earlier refusal to be questioned without counsel in the paradigm Edwards case. That is a ease in which the suspect has been arrested for a particular crime and is held in uninterrupted pretrial custody while that crime is being actively investigated. After the initial interrogation, and up to and including the second one, he remains cut off from his normal life and companions, “thrust into” and isolated in an “unfamiliar,” “police-dominated atmosphere,” Miranda, 384 U. S., at 456-457, where his captors “appear to control [his] fate,” Illinois v. Perkins, 496 U. S. 292, 297 (1990). That was the situation confronted by the suspects in Edwards, Roberson, and Minnick, the three cases in which we have held the Edwards rule applicable. Edwards was arrested pursuant to a warrant and taken to a police station, where he was interrogated until he requested counsel. Edwards, 451 U. S., at 478-479. The officer ended *107the interrogation and took him to the county jail2 but at 9:15 the next morning, two of the officer’s colleagues reinterrogated Edwards at the jail. Id., at 479. Roberson was arrested “at the scene of a just-completed burglary” and interrogated there until he requested a lawyer. Roberson, 486 U. S., at 678. A different officer interrogated him three days later while he “was still in custody pursuant to the arrest.” Ibid. Minnick was arrested by local police and taken to the San Diego jail, where two Federal Bureau of Investigation agents interrogated him the next morning until he requested counsel. Minnick, 498 U. S., at 148-149. Two days later a Mississippi deputy sheriff reinterrogated him at the jail. Id., at 149. None of these suspects regained a sense of control or normalcy after they were initially taken into custody for the crime under investigation.

When, unlike what happened in these three cases, a suspect has been released from his pretrial custody and has returned to his normal life for some time before the later attempted interrogation, there is little reason to think that his change of heart regarding interrogation without counsel has been coerced. He has no longer been isolated. He has likely been able to seek advice from an attorney, family members, and friends.3 And he knows from his earlier experience that he need only demand counsel to bring the interro*108gation to a halt; and that investigative custody does not last indefinitely. In these circumstances, it is farfetched to think that a police officer’s asking the suspect whether he would like to waive his Miranda rights will any more “wear down the accused,” Smith v. Illinois, 469 U. S. 91, 98 (1984) (per curiam), than did the first such request at the original attempted interrogation — which is of course not deemed coercive. His change of heart is less likely attributable to “badgering” than it is to the fact that further deliberation in familiar surroundings has caused him to believe (rightly or wrongly) that cooperating with the investigation is in his interest. Uncritical extension of Edwards to this situation would not significantly increase the number of genuinely coerced confessions excluded. The “justification for a conclusive presumption disappears when application of the presumption will not reach the correct result most of the time.” Coleman v. Thompson, 501 U. S. 722, 737 (1991).

At the same time that extending the Edwards rule yields diminished benefits, extending the rule also increases its costs: the in-fact voluntary confessions it excludes from trial, and the voluntary confessions it deters law enforcement officers from even trying to obtain. Voluntary confessions are not merely “a proper element in law enforcement,” Miranda, supra, at 478, they are an “unmitigated good,” McNeil, 501 U. S., at 181, “ ‘essential to society’s compelling interest in finding, convicting, and punishing those who violate the law,’ ” ibid, (quoting Moran v. Burbine, 475 U. S. 412, 426 (1986)).

The only logical endpoint of Edwards disability is termination of Miranda custody and any of its lingering effects. Without that limitation — -and barring some purely arbitrary time limit4 — every Edwards prohibition of custodial interro*109gation of a particular suspect would be eternal. The prohibition applies, of course, when the subsequent interrogation pertains to a different crime, Roberson, supra, when it is conducted by a different law enforcement authority, Min-nick, 498 U. S. 146, and even when the suspect has met with an attorney after the first interrogation, ibid. And it not only prevents questioning ex ante; it would render invalid, ex post, confessions invited and obtained from suspects who (unbeknownst to the interrogators) have acquired Edwards immunity previously in connection with any offense in any jurisdiction.5 In a country that harbors a large number of repeat offenders,6 this consequence is disastrous.

We conclude that such an extension of Edwards is not justified; we have opened its “'protective umbrella/” Solem, 465 U. S., at 644, n. 4, far enough. The protections offered by Miranda, which we have deemed sufficient to ensure that the police respect the suspect’s desire to have an attorney present the first time police interrogate him, adequately ensure that result when a suspect who initially requested counsel is reinterrogated after a break in custody that is of sufficient duration to dissipate its coercive effects.

*110If Shatzer’s return to the general prison population qualified as a break in custody (a question we address in Part III, infra), there is no doubt that it lasted long enough (two years) to meet that durational requirement. But what about a break that has lasted only one year? Or only one week? It is impractical to leave the answer to that question for clarification in future ease-by-case adjudication;.law enforcement officers need to know, with certainty and beforehand, when renewed interrogation is lawful. And while it is certainly unusual for this Court to set forth precise time limits governing police action, it is not unheard of. In County of Riverside v. McLaughlin, 500 U. S. 44 (1991), we specified 48 hours as the time within which the police must comply with the requirement of Gerstein v. Pugh, 420 U. S. 103 (1975), that a person arrested without a warrant be brought before a magistrate to establish probable cause for continued detention.

Like McLaughlin, this is a case in which the requisite police action (there, presentation to a magistrate; here, abstention from further interrogation) has not been prescribed by statute but has been established by opinion of this Court. We think it appropriate to specify a period of time to avoid the consequence that continuation of the Edwards presumption “will not reach the correct result most of the time.” Coleman, supra, at 737. It seems to us that period is 14 days. That provides plenty of time for the suspect to get reacclimated to his normal life, to consult with friends and counsel, and to shake off any residual coercive effects of his prior custody.

The 14-day limitation meets Shatzer’s concern that a break-in-custody rule lends itself to police abuse. He envisions that once a suspect invokes his Miranda right to counsel, the police will release the suspect briefly (to end the Edwards presumption) and then promptly bring him back into custody for reinterrogation. But once the suspect has been out of custody long enough (14 days) to eliminate its *111coercive effect, there will be nothing to gain by such gamesmanship — nothing, that is, except the entirely appropriate gain of being able to interrogate a suspect who has made a valid waiver of his Miranda rights.7

Shatzer argues that ending the Edwards protections at a break in custody will undermine Edwards’ purpose to conserve judicial resources. To be sure, we have said that “[t]he merit of the Edwards decision lies in the clarity of its command and the certainty of its application.” Minnick, 498 U. S., at 151. But clarity and certainty are not goals in themselves. They are valuable only when they reasonably further the achievement of some substantive end — here, the exclusion of compelled confessions. Confessions obtained after a 2-week break in custody and a waiver of Miranda rights are most unlikely to be compelled, and hence are unreasonably excluded. In any case, a break-in-custody exception will dim only marginally, if at all, the bright-line nature of Edwards. In every case involving Edwards, the courts must determine whether the suspect was in custody when he requested counsel and when he later made the statements he seeks to suppress. Now, in cases where there is an alleged break in custody, they simply have to repeat the inquiry for the time between the initial invocation and reinterrogation. In most cases that determination will be easy. And when it is determined that the defendant pleading Edwards has been out of custody for two weeks before the contested interrogation, the court is spared the fact-intensive *112inquiry into whether he ever, anywhere, asserted his Miranda right to counsel.

III

The facts of this case present an additional issue. No one questions that Shatzer was in custody for Miranda purposes during the interviews with Detective Blankenship in 2003 and Detective Hoover in 2006. Likewise, no one questions that Shatzer triggered the Edwards protections when, according to Detective Blankenship’s notes of the 2003 interview, he stated that “ ‘he would not talk about this case without having an attorney present,’ ” 405 Md., at 589, 954 A. 2d, at 1120. After the 2003 interview, Shatzer was released back into the general prison population where he was serving an unrelated sentence. The issue is whether that constitutes a break in Miranda custody.

We have never decided whether incarceration constitutes custody for Miranda purposes, and have indeed explicitly declined to address the issue. See Perkins, 496 U. S., at 299. See also Bradley v. Ohio, 497 U. S. 1011, 1013 (1990) (Marshall, J., dissenting from denial of certiorari). Whether it does depends upon whether it exerts the coercive pressure that Miranda was designed to guard against — the “danger of coercion [that] results from the interaction of custody and official interrogation.” Perkins, supra, at 297 (emphasis added). To determine whether a suspect was in Miranda custody we have asked whether “there is a ‘formal arrest or restraint on freedom of movement’ of the degree associated with a formal arrest.” New York v. Quarles, 467 U. S. 649, 655 (1984); see also Stansbury v. California, 511 U. S. 318, 322 (1994) (per curiam). This test, no doubt, is satisfied by all forms of incarceration. Our cases make clear, however, that the freedom-of-movement test identifies only a necessary and not a sufficient condition for Miranda custody. We have declined to accord it “talismanic power,” because Miranda is to be enforced “only in those types of situations in which the concerns that powered the decision are impli*113cated.” Berkemer v. McCarty, 468 U. S. 420, 437 (1984). Thus, the temporary and relatively nonthreatening detention involved in a traffic stop or Terry stop, see Terry v. Ohio, 392 U. S. 1 (1968), does not constitute Miranda custody. McCarty, supra, at 439-440. See also Perkins, supra, at 296.

Here, we are addressing the interim period during which a suspect was not interrogated, but was subject to a baseline set of restraints imposed pursuant to a prior conviction. Without minimizing the harsh realities of incarceration, we think lawful imprisonment imposed upon conviction of a crime does not create the coercive pressures identified in Miranda.

Interrogated suspects who have previously been convicted of crime live in prison. When they are released back into the general prison population, they return to their accustomed surroundings and daily routine — they regain the degree of control they had over their lives prior to the interrogation. Sentenced prisoners, in contrast to the Miranda paradigm, are not isolated with their accusers. They live among other inmates, guards, and workers, and often can receive visitors and communicate with people on the outside by mail or telephone.

Their detention, moreover, is relatively disconnected from their prior unwillingness to cooperate in an investigation. The former interrogator has no power to increase the duration of incarceration, which was determined at sentencing.8 And even where the possibility of parole exists, the former interrogator has no apparent power to decrease the time *114served. This is in stark contrast to the circumstances faced by the defendants in Edwards, Roberson, and Minnick, whose continued detention as suspects rested with those controlling their interrogation, and who confronted the uncertainties of what final charges they would face, whether they would be convicted, and what sentence they would receive.

Shatzer’s experience illustrates the vast differences between Miranda custody and incarceration pursuant to conviction. At the time of the 2003 attempted interrogation, Shatzer was already serving a sentence for a prior conviction. After that, he returned to the general prison population in the Maryland Correctional InstitutionHagerstown and was later transferred, for unrelated reasons, down the street to the Roxbury Correctional Institute. Both are medium-security state correctional facilities. See Maryland Div. of Correction Inmate Handbook 7 (2007), online at http://dpscs.md.gov/rehabservs/doc/pdfs/ 2007_Inmate_Handbook.pdf (all Internet materials as visited Feb. 22, 2010, and available in Clerk of Court’s case file). Inmates in these facilities generally can visit the library each week, id., at 28; have regular exercise and recreation periods, id., at 17; can participate in basic adult education and occupational training, id., at 26, 7; are able to send and receive mail, id., at 21-22, 16; and are allowed to receive visitors twice a week, see http://dpscs.md.gov/locations/ mcih.shtml; http://www.dpscs.state.md.us/locations/rci.shtml. His continued detention after the 2003 interrogation did not depend on what he said (or did not say) to Detective Blankenship, and he has not alleged that he was placed in a higher level of security or faced any continuing restraints as a result of the 2003 interrogation. The “inherently compelling pressures” of custodial interrogation ended when he returned to his normal life.

IV

A few words in response to Justice Stevens’ concurrence: It claims we ignore that “[w]hen police tell an indigent *115suspect that he has the right to an attorney” and then “reinterrogate” him without providing a lawyer, “the suspect is likely to feel that the police lied to him and that he really does not have any right to a lawyer.” Post, at 121 (opinion concurring in judgment) (hereinafter concurrence). See also post, at 123, 126, n. 11, 130, n. 16. The fallacy here is that we are not talking about “reinterrogating” the suspect; we are talking about asking his permission to be interrogated. An officer has in no sense lied to a suspect when, after advising, as Miranda requires, ‘You have the right to remain silent, and if you choose to speak you have the right to the presence of an attorney,” he promptly ends the attempted interrogation because the suspect declines to speak without counsel present, and then, two weeks later, reapproaches the suspect and asks, “Are you now willing to speak without a lawyer present?”

The “concer[n] that motivated the Edwards line of cases,” post, at 121, n. 2, is that the suspect will be coerced into saying yes. That concern guides our decision today. Contrary to the concurrence’s conclusion, post, at 122, 124-125, there is no reason to believe a suspect will view confession as “ ‘the only way to end his interrogation’ ” when, before the interrogation begins, he is told that he can avoid it by simply requesting that he not be interrogated without counsel present — an option that worked before. If, as the concurrence argues will often be the case, post, at 124, a break in custody does not change the suspect’s mind, he need only say so.

The concurrence also accuses the Court of “ignor[ing] that when a suspect asks for counsel, until his request is answered, there are still the same ‘inherently compelling’ pressures of custodial interrogation on which the Miranda line of cases is based.” Post, at 123. We do not ignore these pressures; nor do we suggest that they disappear when custody is recommenced after a break, see post, at 124. But if those pressures are merely “the same” as before, then Miranda provides sufficient protection — as it did before. The *116 Edwards presumption of involuntariness is justified only in circumstances where the coercive pressures have increased so much that suspects’ waivers of Miranda rights are likely to be involuntary most of the time. Contrary to the concurrence’s suggestion, post, at 122, it is only in those narrow circumstances — when custody is unbroken — that the Court has concluded a “ ‘fresh se[t] of Miranda warnings’ ” is not sufficient. See Roberson, 486 U. S., at 686.

In the last analysis, it turns out that the concurrence accepts our principal points. It agrees that Edwards prophylaxis is not perpetual; it agrees that a break in custody reduces the inherently compelling pressure upon which Edwards was based; it agrees that Shatzer’s release back into the general prison population constituted a break in custody; and it agrees that in this case the break was long enough to render Edwards inapplicable. Post, at 129-130. We differ in two respects: Instead of terminating Edwards protection when the custodial pressures that were the basis for that protection dissipate, the concurrence would terminate it when the suspect would no longer “feel that he has ‘been denied the counsel he has clearly requested,’” post, at 129. This is entirely unrelated to the rationale of Edwards. If confidence in the police’s promise to provide counsel were the touchstone, Edwards would not have applied in Minnick, where the suspect in continuing custody actually met with appointed counsel. The concurrence’s rule is also entirely unrelated to the existence of a break in custody. While that may relieve the accumulated coercive pressures of custody that are the foundation for Edwards, it is hard to see how it bolsters the suspect’s confidence that if he asks for counsel he will get one.

And secondly, the concurrence differs from us in declining to say how long after a break in custody the termination of Edwards protection occurs. Two and one-half years, it says, is clearly enough — but it gives law enforcement authorities no further guidance. The concurrence criticizes our use of *11714 days as arbitrary and unexplained, post, at 123-124, and n. 7. But in fact that rests upon the same basis as the concurrence’s own approval of a 214-year break in custody: how much time will justify “treating the second interrogation as no more coercive than the first,” post, at 129. Failure to say where the line falls short of 214 years, and leaving that for future case-by-case determination, is certainly less helpful, but not at all less arbitrary.

* * *

Because Shatzer experienced a break in Miranda custody lasting more than two weeks between the first and second attempts at interrogation, Edwards does not mandate suppression of his March 2006 statements. Accordingly, we reverse the judgment of the Court of Appeals of Maryland, and remand the case for further proceedings not inconsistent with this opinion.

It is so ordered.

Justice Thomas,

concurring in part and concurring in the judgment.

I join Part III of the Court’s opinion, which holds that release into the general prison population constitutes a break in custody. I do not join the Court’s decision to extend the presumption of involuntariness established in Edwards v. Arizona, 451 U. S. 477 (1981), for 14 days after custody ends.

It is not apparent to me that the presumption of involuntariness the Court recognized in Edwards is justifiable even in the custodial setting to which Edwards applies it. See, e. g., Minnick v. Mississippi, 498 U. S. 146, 160 (1990) (Scalia, J., dissenting). Accordingly, I would not extend the Edwards rule “beyond the circumstances present in Edwards itself.” 498 U. S., at 162. But even if one believes that the Court is obliged to apply Edwards to any ease involving continuing custody, the Court’s opinion today goes well beyond that. It extends the presumption of involuntar*118iness Edwards applies in custodial settings to interrogations that occur after custody ends.

The Court concedes that this extension, like the Edwards presumption itself, is not constitutionally required. The Court nevertheless defends the extension as a judicially created prophylaxis against compelled confessions. Even if one accepts that such prophylaxis is both permissible generally and advisable for some period following a break in custody,1 the Court's 14-day rule fails to satisfy the criteria our precedents establish for the judicial creation of such a safeguard.

Our precedents insist that judicially created prophylactic rules like those in Edwards and Miranda v. Arizona, 384 U. S. 436 (1966), maintain “the closest possible fit” between the rule and the Fifth Amendment interests they seek to protect. United States v. Patane, 542 U. S. 630, 640-641 (2004) (plurality opinion); see generally Montejo v. Louisiana, 556 U. S. 778, 797 (2009); Chavez v. Martinez, 538 U. S. 760, 772 (2003) (plurality opinion). The Court’s 14-day rule does not satisfy this test. The Court relates its 14-day rule *119to the Fifth Amendment simply by asserting that 14 days between release and recapture should provide “plenty of time for the suspect... to shake off any residual coercive effects of his prior custody,” ante, at 110.

This ipse dixit does not explain why extending the Edwards presumption for 14 days following a break in custody — as opposed to 0,10, or 100 days — provides the “closest possible fit” with the Self-Incrimination Clause, Patane, supra, at 640-641; see ante, at 110 (merely stating that “[i]t seems to us that” the appropriate “period is 14 days”). Nor does it explain how the benefits of a prophylactic 14-day rule (either on its own terms or compared with other possible rules) “outweigh its costs” (which would include the loss of law enforcement information as well as the exclusion of confessions that are in fact voluntary). Ante, at 106 (citing Montejo, supra, at 793).

To be sure, the Court’s rule has the benefit of providing a bright line. Ante, at 111. But bright-line rules are not necessary to prevent Fifth Amendment violations, as the Court has made clear when refusing to adopt such rules in cases involving other Miranda rights. See, e. g., Michigan v. Mosley, 423 U. S. 96, 103-104 (1975). And an otherwise arbitrary rule is not justifiable merely because it gives clear instruction to law enforcement officers.2

As the Court concedes, “clarity and certainty are not goals in themselves. They are valuable only when they reasonably further the achievement of some substantive end — here, the exclusion of compelled confessions” that the Fifth Amendment prohibits. Ante, at 111. The Court’s arbitrary 14-day rule fails this test, even under the relatively permis*120sive criteria set forth in our precedents. Accordingly, I do not join that portion of the Court’s opinion.

Justice Stevens,

concurring in the judgment.

While I agree that the presumption from Edwards v. Arizona, 451 U. S. 477 (1981), is not “eternal,” ante, at 109, and does not mandate suppression of Shatzer’s statement made after a 2-year break in custody, I do not agree with the Court’s newly announced rule: that Edwards always ceases to apply when there is a 14-day break in custody, ante, at 110.

In conducting its “cost-benefit” analysis, the Court demeans Edwards as a “‘second layer’” of “judicially prescribed prophylaxis,” ante, at 104, 105, 111, n. 7; see also ante, at 105 (describing Edwards as “ ‘our rule, not a constitutional command’ ” (quoting Arizona v. Roberson, 486 U. S. 675, 688 (1988) (Kennedy, J., dissenting))). The source of the holdings in the long line of cases that includes both Edwards and Miranda, however, is the Fifth Amendment’s protection against compelled self-incrimination applied to the “compulsion inherent in custodial” interrogation, Miranda v. Arizona, 384 U. S. 436, 458 (1966), and the “significan[ce]” of “the assertion of the right to counsel,” Edwards, 451 U. S., at 485.1 The Court’s analysis today is insufficiently sensitive to the concerns that motivated the Edwards line of cases.

*121I

The most troubling aspect of the Court’s time-based rule is that it disregards the compulsion caused by a second (or third, or fourth) interrogation of an indigent suspect who was told that if he requests a lawyer, one will be provided for him. When police tell an indigent suspect that he has the right to an attorney, that he is not required to speak without an attorney present, and that an attorney will be provided to him at no cost before questioning, the police have made a significant promise. If they cease questioning and then reinterrogate the suspect 14 days later without providing him with a lawyer, the suspect is likely to feel that the police lied to him and that he really does not have any right to a lawyer.2

When officers informed Shatzer of his rights during the first interrogation, they presumably informed him that if he requested an attorney, one would be appointed for him before he was asked any further questions. But if an indigent suspect requests a lawyer, “any further interrogation” (even 14 days later) “without counsel having been provided will surely exacerbate whatever compulsion to speak the suspect may be feeling.” Roberson, 486 U. S., at 686. When police have not honored an earlier commitment to provide a de*122tainee with a lawyer, the detainee likely will “understan[d] his (expressed) wishes to have been ignored” and “may well see further objection as futile and confession (true or not) as the only way to end his interrogation.” Davis v. United States, 512 U. S. 452, 472-473 (1994) (Souter, J., concurring in judgment). Cf. Cooper v. Dupnik, 963 F. 2d 1220, 1225 (CA9 1992) (en banc) (describing an elaborate police task force plan to ignore a suspect’s requests for counsel, on the theory that such would induce hopelessness and thereby elicit an admission). Simply giving a “fresh se[t] of Miranda warnings” will not “‘reassure’ a suspect who has been denied the counsel he has clearly requested that his rights have remained untrammeled.” Roberson, 486 U. S., at 686.

II

The Court never explains why its rule cannot depend on, in addition to a break in custody and passage of time, a concrete event or state of affairs, such as the police’s having honored their commitment to provide counsel. Instead, the Court simply decides to create a time-based rule, and in so doing, disregards much of the analysis upon which Edwards and subsequent decisions were based. “[T]he assertion of the right to counsel” “[i]s a significant event.”3 Edwards, 451 U. S., at 485. As the Court today acknowledges, the *123right to counsel, like the right to remain silent, is one that police may “coere[e] or badge[r],” ante, at 106, a suspect into abandoning.4 However, as discussed above, the Court ignores the effects not of badgering but of reinterrogating a suspect who took the police at their word that he need not answer questions without an attorney present. See Roberson, 486 II. S., at 686. The Court, moreover, ignores that when a suspect asks for counsel, until his request is answered, there are still the same “inherently compelling” pressures of custodial interrogation on which the Miranda line of eases is based, see 486 U. S., at 681,5 and that the concern about compulsion is especially serious for a detainee who has requested a lawyer, an act that signals his “inability to cope with the pressures of custodial interrogation,” id., at 686.6

Instead of deferring to these well-settled understandings of the Edwards rule, the Court engages in its own speeula*124tion that a 14-day break in custody eliminates the compulsion that animated Edwards. But its opinion gives no strong basis for believing that this is the case.7 A 14-day break in custody does not eliminate the rationale for the initial Edwards rule: The detainee has been told that he may remain silent and speak only through a lawyer and that if he cannot afford an attorney, one will be provided for him. He has asked for a lawyer. He does not have one. He is in custody. And police are still questioning him. A 14-day break in custody does not change the fact that custodial interrogation is inherently compelling. It is unlikely to change the fact that a detainee “considers himself unable to deal with the pressures of custodial interrogation without legal assistance.” Roberson, 486 U. S., at 683.8 And in some instances, a 14-day break in custody may make matters worse9 “[w]hen a *125suspect understands his (expressed.) wishes to have been ignored” and thus “may well see further objection as futile and confession (true or not) as the only way to end his interrogation.” Davis, 512 U. S., at 472-473 (Souter, J., concurring in judgment).10

The Court ignores these understandings from the Edwards line of cases and instead speculates that if a suspect is reinterrogated and eventually talks, it must be that “further deliberation in familiar surroundings has caused him to believe (rightly or wrongly) that cooperating with the investigation is in his interest.” Ante, at 108. But it is not apparent why that is the case. The answer, we are told, is that once a suspect has been out of Miranda custody for 14 days, “[h]e has likely been able to seek advice from an attorney, family members, and Mends.” Ante, at 107. This speculation, however, is overconfident and only questionably relevant. As a factual matter, we do not know whether the defendant has been able to seek advice: First of all, suspects are told that if they cannot afford a lawyer, one will be provided for them. Yet under the majority’s rule, an indigent suspect who took the police at their word when he asked for a lawyer will nonetheless be assumed to have “been able to seek advice from an attorney.” Second, even suspects who *126are not indigent cannot necessarily access legal advice (or social advice as the Court presumes) within 14 days. Third, suspects may not realize that they need to seek advice from an attorney. Unless police warn suspects that the interrogation will resume in 14 days, why contact a lawyer? When a suspect is let go, he may assume that the police were satisfied. In any event, it is not apparent why interim advice matters.11 In Minnick v. Mississippi, 498 U. S. 146, 153 (1990), we held that it is not sufficient that a detainee happened to speak at some point with a lawyer. See ibid, (noting that “consultation with an attorney” does not prevent “persistent attempts by officials to persuade [a suspect] to waive his rights” or shield against the “coercive pressures that accompany custody”). If the actual interim advice of an attorney is not sufficient, the hypothetical, interim advice of “an attorney, family members, and friends,” ante, at 107, is not enough.

The many problems with the Court’s new rule are exacerbated in the very situation in this case: a suspect who is in prison. Even if, as the Court assumes, a trip to one’s home significantly changes the Edwards calculus, a trip to one’s prison cell is not the same. A prisoner’s freedom is severely limited, and his entire life remains subject to government control. Such an environment is not conducive to “shak[ing] off any residual coercive effects of his prior custody.” Ante, at 110.12 Nor can a prisoner easily “seek advice from an at*127torney, family members, and friends,” ante, at 107, especially not within 14 days; prisoners are frequently subject to restrictions on communications. Nor, in most cases, can he live comfortably knowing that he cannot be badgered by police; prison is not like a normal situation in which a suspect “is in control, and need only shut his door or walk away to avoid police badgering.” Montejo v. Louisiana, 556 U. S. 778, 795 (2009). Indeed, for a person whose every move is controlled by the State, it is likely that “his sense of dependence on, and trust in, counsel as the guardian of his interests in dealing with government officials intensified.” United States v. Green, 592 A. 2d 985, 989 (D. C. 1991); cf. Minnick, 498 U. S., at 153 (explaining that coercive pressures “may increase as custody is prolonged”).13 The Court ignores these realities of prison, and instead rests its argument on the supposition that a prisoner's “detention ... is relatively disconnected from their prior unwillingness to cooperate in an investigation.” Ante, at 113. But that is not necessarily the case. Prisoners are uniquely vulnerable to the officials who control every aspect of their lives; prison guards may not look kindly upon a prisoner who refuses to cooperate with police. And cooperation frequently is relevant to *128whether the prisoner can obtain parole. See, e. g., Code of Md. Regs., tit. 12, § 08.01.18(A)(3) (2008). Moreover, even if it is true as a factual matter that a prisoner’s fate is not controlled by the police who come to interrogate him, how is the prisoner supposed to know that? As the Court itself admits, compulsion is likely when a suspect’s “captors appear to control [his] fate,” ante, at 106 (internal quotation marks omitted). But when a guard informs a suspect that he must go speak with police, it will “appear” to the prisoner that the guard and police are not independent. “Questioning by captors, who appear to control the suspect’s fate, may create mutually reinforcing pressures that the Court has assumed will weaken the suspect’s will.” Illinois v. Perkins, 496 U. S. 292, 297 (1990) (emphasis added).14

*129III

Because, at the very least, we do not know whether Shatzer could obtain a lawyer, and thus would have felt that police had lied about providing one, I cannot join the Court’s opinion. I concur in today’s judgment, however, on another ground: Even if Shatzer eould not consult a lawyer and the police never provided him one, the 2-year break in custody is a basis for treating the second interrogation as no more coercive than the first. Neither a break in custody nor the passage of time has an inherent, curative power. But certain things change over time. An indigent suspect who took police at their word that they would provide an attorney probably will feel that he has “been denied the counsel he has clearly requested,” Roberson, 486 U. S., at 686, when police begin to question him, without a lawyer, only 14 days later.15 But, when a suspect has been left alone for a sig*130nificant period of time, he is not as likely to draw such conclusions when the police interrogate him again.16 It is concededly “impossible to determine with precision” where to draw such a line. Barker v. Wingo, 407 U. S. 514, 521 (1972). In the case before us, however, the suspect was returned to the general prison population for two years. I am convinced that this period of time is sufficient. I therefore concur in the judgment.

*415Justice O’Connor

delivered the opinion of the Court.

After being informed of his rights pursuant to Miranda v. Arizona, 384 U. S. 436 (1966), and after executing a series of written waivers, respondent confessed to the murder of a young woman. At no point during the course of the interrogation, which occurred prior to arraignment, did he request an attorney. While he was in police custody, his sister attempted to retain a lawyer to represent him. The attorney telephoned the police station and received assurances that respondent would not be questioned further until the next day. In fact, the interrogation session that yielded the inculpatory statements began later that evening. The question presented is whether either the conduct of the police or respond*416ent’s ignorance of the attorney’s efforts to reach him taints the validity of the waivers and therefore requires exclusion of the confessions.

I

On the morning of March 3, 1977, Mary Jo Hickey was found unconscious in a factory parking lot in Providence, Rhode Island. Suffering from injuries to her skull apparently inflicted by a metal pipe found at the scene, she was rushed to a nearby hospital. Three weeks later she died from her wounds.

Several months after her death, the Cranston, Rhode Island, police arrested respondent and two others in connection with a local burglary. Shortly before the arrest, Detective Ferranti of the Cranston police force had learned from a confidential informant that the man responsible for Ms. Hickey’s death lived at a certain address and went by the name of “Butch. ” Upon discovering that respondent lived at that address and was known by that name, Detective Ferranti informed respondent of his Miranda rights. When respondent refused to execute a written waiver, Detective Ferranti spoke separately with the two other suspects arrested on the breaking and entering charge and obtained statements further implicating respondent in Ms. Hickey’s murder. At approximately 6 p.m., Detective Ferranti telephoned the police in Providence to convey the information he had uncovered. An hour later, three officers from that department arrived at the Cranston headquarters for the purpose of questioning respondent about the murder.

That same evening, at about 7:45 p.m., respondent’s sister telephoned the Public Defender’s Office to obtain legal assistance for her brother. Her sole concern was the breaking and entering charge, as she was unaware that respondent was then under suspicion for murder. She asked for Richard Casparian who had been scheduled to meet with respondent earlier that afternoon to discuss another charge unrelated to either the break-in or the murder. As soon as the conversa*417tion ended, the attorney who took the call attempted to reach Mr. Casparian. When those efforts were unsuccessful, she telephoned Allegra Munson, another Assistant Public Defender, and told her about respondent’s arrest and his sister’s subsequent request that the office represent him.

At 8:15 p.m., Ms. Munson telephoned the Cranston police station and asked that her call be transferred to the detective division. In the words of the Supreme Court of Rhode Island, whose factual findings we treat as presumptively correct, 28 U. S. C. § 2254(d), the conversation proceeded as follows:

“A male voice responded with the word ‘Detectives.’ Ms. Munson identified herself and asked if Brian Burbine was being held; the person responded affirmatively. Ms. Munson explained to the person that Burbine was represented by attorney Casparian who was not available; she further stated that she would act as Burbine’s legal counsel in the event that the police intended to place him in a lineup or question him. The unidentified person told Ms. Munson that the police would not be questioning Burbine or putting him in a lineup and that they were through with him for the night. Ms. Munson was not informed that the Providence Police were at the Cranston police station or that Burbine was a suspect in Mary’s murder.” State v. Burbine, 451 A. 2d 22, 23-24 (1982).

At all relevant times, respondent was unaware of his sister’s efforts to retain counsel and of the fact and contents of Ms. Munson’s telephone conversation.

Less than an hour later, the police brought respondent to an interrogation room and conducted the first of a series of interviews concerning the murder. Prior to each session, respondent was informed of his Miranda rights, and on three separate occasions he signed a written form acknowledging that he understood his right to the presence of an attorney and explicitly indicating that he “[did] not want an attorney *418called or appointed for [him]” before he gave a statement. App. to Pet. for Cert. 94, 103, 107. Uncontradicted evidence at the suppression hearing indicated that at least twice during the course of the evening, respondent was left in a room where he had access to a telephone, which he apparently declined to use. Tr. of Suppression Hearing 23, 85. Eventually, respondent signed three written statements fully admitting to the murder.

Prior to trial, respondent moved to suppress the statements. The court denied the motion, finding that respondent had received the Miranda warnings and had “knowingly, intelligently, and voluntarily waived his privilege against self-incrimination [and] his right to counsel.” App. to Pet. for Cert. 116. Rejecting the contrary testimony of the police, the court found that Ms. Munson did telephone the detective bureau on the evening in question, but concluded that “there was no . . . conspiracy or collusion on the part of the Cranston Police Department to secrete this defendant from his attorney.” Id., at 114. In any event, the court held, the constitutional right to request the presence of an attorney belongs solely to the defendant and may not be asserted by his lawyer. Because the evidence was clear that respondent never asked for the services of an attorney, the telephone call had no relevance to the validity of the waiver or the admissibility of the statements.

The jury found respondent guilty of murder in the first degree, and he appealed to the Supreme Court of Rhode Island. A divided court rejected his contention that the Fifth and Fourteenth Amendments to the Constitution required , the suppression of the inculpatory statements and affirmed the conviction. Failure to inform respondent of Ms. Munson’s efforts to represent him, the court held, did not undermine the validity of the waivers. “It hardly seems conceivable that the additional information that an attorney whom he did not know had called the police station would have added significantly to the quantum of information necessary for the *419accused to make an informed decision as to waiver. ” State v. Burbine 451 A. 2d 22, 29 (1982). Nor, the court concluded, did Miranda v. Arizona, or any other decision of this Court independently require the police to honor Ms. Munson’s request that interrogation not proceed in her absence. In reaching that conclusion, the court noted that because two different police departments were operating in the Cranston station house on the evening in question, the record supported the trial court’s finding that there was no “conspiracy or collusion” to prevent Ms. Munson from seeing respondent. 451 A. 2d, at 30, n. 5. In any case, the court held, the right to the presence of counsel belongs solely to the accused and may not be asserted by “benign third parties, whether or not they happen to be attorneys.” Id., at 28.

After unsuccessfully petitioning the United States District Court for the District of Rhode Island for a writ of habeas corpus, 589 F. Supp. 1245 (1984), respondent appealed to the Court of Appeals for the First Circuit. That court reversed. 753 F. 2d 178 (1985). Finding it unnecessary to reach any arguments under the Sixth and Fourteenth Amendments, the court held that the police’s conduct had fatally tainted respondent’s “otherwise valid” waiver of his Fifth Amendment privilege against self-incrimination and right to counsel. Id., at 184. The court reasoned that by failing to inform respondent that an attorney had called and that she had been assured that no questioning would take place until the next day, the police had deprived respondent of information crucial to his ability to waive his rights knowingly and intelligently. The court also found that the record would support “no other explanation for the refusal to tell Burbine of Attorney Munson’s call than . . . deliberate or reckless irresponsibility.” Id., at 185. This kind of “blameworthy action by the police,” the court concluded, together with respondent’s ignorance of the telephone call, “vitiate[d] any claim that [the] waiver of counsel was knowing and voluntary.” Id., at 185, 187.

*420We granted certiorari to decide whether a prearraignment confession preceded by an otherwise valid waiver must be suppressed either because the police misinformed an inquiring attorney about their plans concerning the suspect or because they failed to inform the suspect of the attorney’s efforts to reach him. 471 U. S. 1098 (1985). We now reverse.

r — I ⅜ — I

In Miranda v. Arizona, the Court recognized that custodial interrogations, by their very nature, generate “compelling pressures which work to undermine the individual’s will to resist and to compel him to speak where he would not otherwise do so freely.” 384 U. S., at 467. To combat this inherent compulsion, and thereby protect the Fifth Amendment privilege against self-incrimination, Miranda imposed on the police an obligation to follow certain procedures in their dealings with the accused. In particular, prior to the initiation of questioning, they must fully apprise the suspect of the State’s intention to use his statements to secure a conviction, and must inform him of his rights to remain silent and to “have counsel present ... if [he] so desires.” Id., at 468-470. Beyond this duty to inform, Miranda requires that the police respect the accused’s decision to exercise the rights outlined in the warnings. “If the individual indicates in any manner, at any time prior to or during questioning, that he wishes to remain silent, [or if he] states that he wants an attorney, the interrogation must cease.” Id., at 473-474. See also Edwards v. Arizona, 451 U. S. 477 (1981).

Respondent does not dispute that the Providence police followed these procedures with precision. The record amply supports the state-court findings that the police administered the required warnings, sought to assure that respondent understood his rights, and obtained an express written waiver prior to eliciting each of the three statements. Nor does respondent contest the Rhode Island courts’ determination that he at no point requested the presence of a lawyer. *421He contends instead that the confessions must be suppressed because the police’s failure to inform him of the attorney’s telephone call deprived him of information essential to his ability to knowingly waive his Fifth Amendment rights. In the alternative, he suggests that to fully protect the Fifth Amendment values served by Miranda, we should extend that decision to condemn the conduct of the Providence police. We address each contention in turn.

A

Echoing the standard first articulated in Johnson v. Zerbst, 304 U. S. 458, 464 (1938), Miranda holds that “[t]he defendant may waive effectuation” of the rights conveyed in the warnings “provided the waiver is made voluntarily, knowingly and intelligently.” 384 U. S., at 444, 475. The inquiry has two distinct dimensions. Edwards v. Arizona, supra, at 482; Brewer v. Williams, 430 U. S. 387, 404 (1977). First, the relinquishment of the right must have been voluntary in the sense that it was the product of a free and deliberate choice rather than intimidation, coercion, or deception. Second, the waiver must have been made with a full awareness of both the nature of the right being abandoned and the consequences of the decision to abandon it. Only if the “totality of the circumstances surrounding the interrogation” reveals both an uncoerced choice and the requisite level of comprehension may a court properly conclude that the Miranda rights have been waived. Fare v. Michael C., 442 U. S. 707, 725 (1979). See also North Carolina v. Butler, 441 U. S. 369, 374-375 (1979).

Under this standard, we have no doubt that respondent validly waived his right to remain silent and to the presence of counsel. The voluntariness of the waiver is not at issue. As the Court of Appeals correctly acknowledged, the record is devoid of any suggestion that police resorted to physical or psychological pressure to elicit the statements. 753 F. 2d, at 184. Indeed it appears that it was respondent, and not the *422police, who spontaneously initiated the conversation that led to the first and most damaging confession. Id., at 180. Cf. Edwards v. Arizona, supra. Nor is there any question about respondent’s comprehension of the full panoply of rights set out in the Miranda warnings and of the potential consequences of a decision to relinquish them. Nonetheless, the Court of Appeals believed that the “[deliberate or reckless” conduct of the police, in particular their failure to inform respondent of the telephone call, fatally undermined the validity of the otherwise proper waiver. 753 F. 2d, at 187. We find this conclusion untenable as a matter of both logic and precedent.

Events occurring outside of the presence of the suspect and entirely unknown to him surely can have no bearing on the capacity to comprehend and knowingly relinquish a constitutional right. Under the analysis of the Court of Appeals, the same defendant, armed with the same information and confronted with precisely the same police conduct, would have knowingly waived his Miranda rights had a lawyer not telephoned the police station to inquire about his status. Nothing in any of our waiver decisions or in our understanding of the essential components of a valid waiver requires so incongruous a result. No doubt the additional information would have been useful to respondent; perhaps even it might have affected his decision to confess. But we have never read the Constitution to require that the police supply a suspect with a flow of information to help him calibrate his self-interest in deciding whether to speak or stand by his rights. See, e. g., Oregon v. Elstad, 470 U. S. 298, 316-317 (1985); United States v. Washington, 431 U. S. 181, 188 (1977). Cf. Hill v. Lockhart, 474 U. S. 52, 56 (1985); McMann v. Richardson, 397 U. S. 759, 769 (1970). Once it is determined that a suspect’s decision not to rely on his rights was uncoerced, that he at all times knew he could stand mute and request a lawyer, and that he was aware of the State’s intention to use his statements to secure a conviction, the analysis *423is complete and the waiver is valid as a matter of law.1 The Court of Appeals’ conclusion to the contrary was in error.

Nor do we believe that the level of the police’s culpability in failing to inform respondent of the telephone call has any bearing on the validity of the waivers. In light of the state-court findings that there was no “conspiracy or collusion” on the part of the police, 451 A. 2d, at 30, n. 5, we have serious doubts about whether the Court of Appeals was free to conclude that their conduct constituted “deliberate or reckless irresponsibility.” 753 F. 2d, at 185; see 28 U. S. C. § 2254(d). But whether intentional or inadvertent, the state of mind of the police is irrelevant to the question of the intelligence and voluntariness of respondent’s election to abandon his rights. Although highly inappropriate, even deliberate deception of an attorney could not possibly affect a suspect’s decision to waive his Miranda rights unless he were at least aware of the incident. Compare Escobedo v. Illinois, 378 U. S. 478, 481 (1964) (excluding confession where police incorrectly told the suspect that his lawyer “ ‘didn’t want to see’ him”). Nor was the failure to inform respondent of the telephone call the kind of “trick[ery]” that can vitiate the validity of a waiver. Miranda, 384 U. S., at 476. Granting that the “deliberate or reckless” withholding of information is objectionable as a *424matter of ethics, such conduct is only relevant to the constitutional validity of a waiver if it deprives a defendant of knowledge essential to his ability to understand the nature of his rights and the consequences of abandoning them. Because respondent’s voluntary decision to speak was made with full awareness and comprehension of all the information Miranda requires the police to convey, the waivers were valid.

B

At oral argument respondent acknowledged that a constitutional rule requiring the police to inform a suspect of an attorney’s efforts to reach him would represent a significant extension of our precedents. Tr. of Oral Arg. 32-33. He contends, however, that the conduct of the Providence police was so inimical to the Fifth Amendment values Miranda seeks to protect that we should read that decision to condemn their behavior. Regardless of any issue of waiver, he urges, the Fifth Amendment requires the reversal of a conviction if the police are less than forthright in their dealings with an attorney or if they fail to tell a suspect of a lawyer’s unilateral efforts to contact him. Because the proposed modification ignores the underlying purposes of the Miranda rules and because we think that the decision as written strikes the proper balance between society’s legitimate law enforcement interests and the protection of the defendant’s Fifth Amendment rights, we decline the invitation to further extend Miranda’s reach.

At the outset, while we share respondent’s distaste for the deliberate misleading of an officer of the court, reading Miranda to forbid police deception of an attorney “would cut [the decision] completely loose from its own explicitly stated rationale.” Beckwith v. United States, 425 U. S. 341, 345 (1976). As is now well established, “[t]he . . . Miranda warnings are ‘not themselves rights protected by the Constitution but [are] instead measures to insure that the [suspect’s] right against compulsory self-incrimination [is] pro*425tected.’” New York v. Quarles, 467 U. S. 649, 654 (1984), quoting Michigan v. Tucker, 417 U. S. 433, 444 (1974). Their objective is not to mold police conduct for its own sake. Nothing in the Constitution vests in us the authority to mandate a code of behavior for state officials wholly unconnected to any federal right or privilege. The purpose of the Miranda warnings instead is to dissipate the compulsion inherent in custodial interrogation and, in so doing, guard against abridgment of the suspect’s Fifth Amendment rights. Clearly, a rule that focuses on how the police treat an attorney-conduct that has no relevance at all to the degree of compulsion experienced by the defendant during interrogation — would ignore both Miranda’s mission and its only source of legitimacy.

Nor are we prepared to adopt a rule requiring that the police inform a suspect of an attorney’s efforts to reach him. While such a rule might add marginally to Miranda’s goal of dispelling the compulsion inherent in custodial interrogation, overriding practical considerations counsel against its adoption. As we have stressed on numerous occasions, “[o]ne of the principal advantages” of Miranda is the ease and clarity of its application. Berkemer v. McCarty, 468 U. S. 420, 430 (1984); see also New York v. Quarles, supra, at 662-664 (concurring opinion); Fare v. Michael C., 442 U. S., at 718. We have little doubt that the approach urged by respondent and endorsed by the Court of Appeals would have the inevitable consequence of muddying Miranda’s otherwise relatively clear waters. The legal questions it would spawn are legion: To what extent should the police be held accountable for knowing that the accused has counsel? Is it enough that someone in the station house knows, or must the interrogating officer himself know of counsel’s efforts to contact the suspect? Do counsel’s efforts to talk to the suspect concerning one criminal investigation trigger the obligation to inform the defendant before interrogation may proceed on a wholly separate matter? We are unwilling to modify Miranda in a *426manner that would so clearly undermine the decision’s central “virtue of informing police and prosecutors with specificity .. . what they may do in conducting [a] custodial interrogation, and of informing courts under what circumstances statements obtained during such interrogation are not admissible.” Fare v. Michael C., supra, at 718.

Moreover, problems of clarity to one side, reading Miranda to require the police in each instance to inform a suspect of an attorney’s efforts to reach him would work a substantial and, we think, inappropriate shift in the subtle balance struck in that decision. Custodial interrogations implicate two competing concerns. On the one hand, “the need for police questioning as a tool for effective enforcement of criminal laws” cannot be doubted. Schneckloth v. Bustamonte, 412 U. S. 218, 225 (1973). Admissions of guilt are more than merely “desirable,” United States v. Washington, 431 U. S., at 186; they are essential to society’s compelling interest in finding, convicting, and punishing those who violate the law. On the other hand, the Court has recognized that the interrogation process is “inherently coercive” and that, as a consequence, there exists a substantial risk that the police will inadvertently traverse the fine line between legitimate efforts to elicit admissions and constitutionally impermissible compulsion. New York v. Quarles, supra, at 656. Miranda attempted to reconcile these opposing concerns by giving the defendant the power to exert some control over the course of the interrogation. Declining to adopt the more extreme position that the actual presence of a lawyer was necessary to dispel the coercion inherent in custodial interrogation, see Brief for American Civil Liberties Union as Amicus Curiae in Miranda v. Arizona, O. T. 1965, No. 759, pp. 22-31, the Court found that the suspect’s Fifth Amendment rights could be adequately protected by less intrusive means. Police questioning, often an essential part of the investigatory process, could continue in its traditional form, the Court held, but only if the suspect clearly under*427stood that, at any time, he could bring the proceeding to a halt or, short of that, call in an attorney to give advice and monitor the conduct of his interrogators.

The position urged by respondent would upset this carefully drawn approach in a manner that is both unnecessary for the protection of the Fifth Amendent privilege and injurious to legitimate law enforcement. Because, as Miranda holds, full comprehension of the rights to remain silent and request an attorney are sufficient to dispel whatever coercion is inherent in the interrogation process, a rule requiring the police to inform the suspect of an attorney’s efforts to contact him would contribute to the protection of the Fifth Amendment privilege only incidentally, if at all. XhisJxdnimal benefit, however, would come at a substantial cost to society’s legitimate and substantial interest in securing admissions of guilt. Indeed, the very premise of the Court of Appeals was not that awareness of Ms. Munson’s phone call would have dissipated the coercion of the interrogation room, but that it might have convinced respondent not to speak at all. 753 F. 2d, at 185. Because neither the letter nor purposes of Miranda require this additional handicap on otherwise permissible investigatory efforts, we are unwilling to expand the Miranda rules to require the police to keep the suspect abreast of the status of his legal representation.

We acknowledge that a number of state courts have reached a contrary conclusion. Compare State v. Jones, 19 Wash. App. 850, 578 P. 2d 71 (1978), with State v. Beck, 687 S. W. 2d 155 (Mo. 1985) (en banc). We recognize also that our interpretation of the Federal Constitution, if given the dissent’s expansive gloss, is at odds with the policy recommendations embodied in the American Bar Association Standards of Criminal Justice. Cf. ABA Standards for Criminal Justice 5-7.1 (2d ed. 1980). Notwithstanding the dissent’s protestations, however, our interpretive duties go well beyond deferring to the numerical preponderance of lower court decisions or to the subconstitutional recommen*428dations of even so esteemed a body as the American Bar Association. See Nix v. Whiteside, ante, at 189 (Blackmun, J., concurring in judgment). Nothing we say today disables the States from adopting different requirements for the conduct of its employees and officials as a matter of state law. We hold only that the Court of Appeals erred in construing the Fifth Amendment to the Federal Constitution to require the exclusion of respondent’s three confessions.

I — I I — I I — I

Respondent also contends that the Sixth Amendment requires exclusion of his three confessions.2 It is clear, of course, that, absent a valid waiver, the defendant has the right to the presence of an attorney during any interrogation occurring after the first formal charging proceeding, the point at which the Sixth Amendment right to counsel initially attaches. United States v. Gouveia, 467 U. S. 180, 187 (1984); Kirby v. Illinois, 406 U. S. 682, 689 (1972) (opinion of Stewart, J.). See Brewer v. Williams, 430 U. S., at 400-401. And we readily agree that once the right has attached, it follows that the police may not interfere with the efforts of a defendant’s attorney to act as a “‘medium’ between [the suspect] and the State” during the interrogation. Maine v. Moulton, 474 U. S. 159, 176 (1985); see Brewer v. Williams, supra, at 401, n. 8. The difficulty for respondent is that the interrogation sessions that yielded the inculpatory statements took place before the initiation of “adversary judicial proceedings.” United States v. Gouveia, supra, at 192. He contends, however, that this circumstance is not fatal to his Sixth Amendment claim. At least in some situations, he argues, the Sixth Amendment protects the integrity of the *429attorney-client relationship3 regardless of whether the prosecution has in fact commenced “by way of formal charge, preliminary hearing, indictment, information or arraignment.” 467 U. S., at 188. Placing principal reliance on a footnote in Miranda, 384 U. S., at 465, n. 35, and on Escobedo v. Illinois, 378 U. S. 478 (1964), he maintains that Gouveia, Kirby, and our other “critical stage” cases, concern only the narrow question of when the right to counsel — that is, to the appointment or presence of counsel — attaches. The right to noninterference with an attorney’s dealings with a criminal suspect, he asserts, arises the moment that the relationship is formed, or, at the very least, once the defendant is placed in custodial interrogation.

We are not persuaded. At the outset, subsequent decisions foreclose any reliance on Escobedo and Miranda for the proposition that the Sixth Amendment right, in any of its manifestations, applies prior to the initiation of adversary judicial proceedings. Although Escobedo was originally decided as a Sixth Amendment case, “the Court in retrospect perceived that the ‘prime purpose’ of Escobedo was not to vindicate the constitutional right to counsel as such, but, like Miranda, ‘to guarantee full effectuation of the privilege against self-incrimination . . . .’” Kirby v. Illinois, supra, *430at 689, quoting Johnson v. New Jersey, 384 U. S. 719, 729 (1966), Clearly then, Escobedo provides no support for respondent’s argument. Nor, of course, does Miranda, the holding of which rested exclusively on the Fifth Amendment. Thus, the decision’s brief observation about the reach of Escobedo’s Sixth Amendment analysis is not only dictum, but reflects an understanding of the case that the Court has expressly disavowed. See also, United States v. Gouveia, supra, at 188, n. 5; Y. Kamisar, Police Interrogation and Confessions 217-218, n. 94 (1980).

Questions of precedent to one side, we find respondent’s understanding of the Sixth Amendment both practically and theoretically unsound. As a practical matter, it makes little sense to say that the Sixth Amendment right to counsel attaches at different times depending on the fortuity of whether the suspect or his family happens to have retained counsel prior to interrogation. Cf. id., at 220-221. More importantly, the suggestion that the existence of an attorney-client relationship itself triggers the protections of the Sixth Amendment misconceives the underlying purposes of the right to counsel. The Sixth Amendment’s intended function is not to wrap a protective cloak around the attorney-client relationship for its own sake any more than it is to protect a suspect from the consequences of his own candor. Its purpose, rather, is to assure that in any “criminal prosecutio[n],” U. S. Const., Amdt. 6, the accused shall not be left to his own devices in facing the “ ‘prosecutorial forces of organized society.’” Maine v. Moulton, supra, at 170 (quoting Kirby v. Illinois, 406 U. S., at 689). By its very terms, it becomes applicable only when the government’s role shifts from investigation to accusation. For it is only then that the assistance of one versed in the “intricacies ... of law,” ibid., is needed to assure that the prosecution’s case encounters “the crucible of meaningful adversarial testing.” United States v. Cronic, 466 U. S. 648, 656 (1984).

*431Indeed, in Maine v. Moulton, decided this Term, the Court again confirmed that looking to the initiation of adversary judicial proceedings, far from being mere formalism, is fundamental to the proper application of the Sixth Amendment right to counsel. There, we considered the constitutional implications of a surreptitious investigation that yielded evidence pertaining to two crimes. For one, the defendant had been indicted; for the other, he had not. Concerning the former, the Court reaffirmed that after the first charging proceeding the government may not deliberately elicit incriminating statements from an accused out of the presence of counsel. See also Massiah v. United States, 377 U. S. 201 (1964). The Court made clear, however, that the evidence concerning the crime for which the defendant had not been indicted — evidence obtained in precisely the same manner from the identical suspect — would be admissible at a trial limited to those charges. Maine v. Moulton, 474 U. S., at 180, and n. 16. The clear implication of the holding, and one that confirms the teaching of Gouveia, is that the Sixth Amendment right to counsel does not attach until after the initiation of formal charges. Moreover, because Moulton already had legal representation, the decision all but forecloses respondent’s argument that the attorney-client relationship itself triggers the Sixth Amendment right.

Respondent contends, however, that custodial interrogations require a different rule. Because confessions elicited during the course of police questioning often seal a suspect’s fate, he argues, the need for an advocate — and the concomitant right to noninterference with the attorney-client relationship — is at its zenith, regardless of whether the State has initiated the first adversary judicial proceeding. We do not doubt that a lawyer’s presence could be of value to the suspect; and we readily agree that if a suspect confesses, his attorney’s case at trial will be that much more difficult. But these concerns are no more decisive in this context than they were for the equally damaging preindictment lineup *432at issue in Kirby, or the statements pertaining to the unindicted crime elicted from the defendant in Maine v. Moulton. Compare United States v. Wade, 388 U. S. 218, 226-227 (1967) (Sixth Amendment attaches at postindictment lineup); Massiah v. United States, supra (after indictment, police may not elicit statements from suspect out of the presence of counsel). For an interrogation, no more or less than for any other “critical” pretrial event, the possibility that the encounter may have important consequences at trial, standing alone, is insufficient to trigger the Sixth Amendment right to counsel. As Gouveia made clear, until such time as the “ ‘government has committed itself to prosecute, and . . . the adverse positions of government and defendant have solidified’” the Sixth Amendment right to counsel does not attach. 467 U. S., at 189 (quoting Kirby v. Illinois, supra, at 689).

Because, as respondent acknowledges, the events that led to the inculpatory statements preceded the formal initiation of adversary judicial proceedings, we reject the contention that the conduct of the police violated his rights under the Sixth Amendment.

IV

Finally, respondent contends that the conduct of the police was so offensive as to deprive him of the fundamental fairness guaranteed by the Due Process Clause of the Fourteenth Amendment. Focusing primarily on the impropriety of conveying false information to an attorney, he invites us to declare that such behavior should be condemned as violative of canons fundamental to the “‘traditions and conscience of our people.’” Rochin v. California, 342 U. S. 165, 169 (1952), quoting Snyder v. Massachusetts, 291 U. S. 97, 105 (1934). We do not question that on facts more egregious than those presented here police deception might rise to a level of a due process violation. Accordingly, Justice Ste-*433VENS’ apocalyptic suggestion that we have approved any and all forms of police misconduct is demonstrably incorrect.4 We hold only that, on these facts, the challenged conduct falls short of the kind of misbehavior that so shocks the sensibil*434ities of civilized society as to warrant a federal intrusion into the criminal processes of the States.

We hold therefore that the Court of Appeals erred in finding that the Federal Constitution required the exclusion of the three inculpatory statements. Accordingly, we reverse and remand for proceedings consistent with this opinion.

So ordered.

Justice Stevens,

with whom Justice Brennan and Justice Marshall join, dissenting.

This case poses fundamental questions about our system of justice. As this Court has long recognized, and reaffirmed only weeks ago, “ours is an accusatorial and not an inquisitorial system.” Miller v. Fenton, 474 U. S. 104, 110 (1985).1 The Court’s opinion today represents a startling departure from that basic insight.

*435The Court concludes that the police may deceive an attorney by giving her false information about whether her client will be questioned, and that the police may deceive a suspect by failing to inform him of his attorney’s communications and efforts to represent him.2 For the majority, this conclusion, though “distaste[ful],” ante, at 424, is not even debatable. The deception of the attorney is irrelevant because the attorney has no right to information, accuracy, honesty, or fairness in the police response to her questions about her client. The deception of the client is acceptable, because, although the information would affect the client’s assertion of his rights, the client’s actions in ignorance of the availability of his attorney are voluntary, knowing, and intelligent; additionally, society’s interest in apprehending, prosecuting, and punishing criminals outweighs the suspect’s interest in information regarding his attorney’s efforts to communicate with him. Finally, even mendacious police interference in the communications between a suspect and his lawyer does not violate any notion of fundamental fairness because it does not shock the conscience of the majority.

The case began in March 1977 with the discovery of Mary Jo Hickey, unconscious and disheveled in a deserted parking lot, lying in a pool of blood, with semen on her clothes, her dentures broken, and a piece of heavy, bloodstained metal nearby. Days later, Brian Burbine, then 20 years old, went to Maine and stayed with friends. According to the friends’ testimony at trial, he was upset, and described a night out with Hickey, who was then 35. After several drinks, *436Burbine told them, a ride home turned into a violent encounter; he hit Hickey several times and threw her out of the car. Three weeks after she was discovered in the parking lot, Hickey died. Three months later, after the 21-hour period of detention by the Cranston and Providence, Rhode Island, police that is the focus of this dispute, Burbine was charged with her murder, and ultimately found guilty of it.

The murder of Mary Jo Hickey was a vicious crime, fully meriting a sense of outrage and a desire to find and prosecute the perpetrator swiftly and effectively. Indeed, by the time Burbine was arrested on an unrelated breaking-and-entering charge, the Hickey murder had been the subject of a local television special.3 Not surprisingly, Detective Ferranti, the Cranston Detective who “broke” the case, was rewarded with a special commendation for his efforts.4

The recognition that ours is an accusatorial, and not an inquisitorial system nevertheless requires that the government’s actions, even in responding to this brutal crime, respect those liberties and rights that distinguish this society from most others. As Justice Jackson observed shortly after his return from Nuremberg, cases of this kind present “a real dilemma in a free society ... for the defendant is shielded by such safeguards as no system of law except the Anglo-American concedes to him.”5 Justice Frankfurter similarly *437emphasized that it is “a fair summary of history to say that the safeguards of liberty have been forged in controversies involving not very nice people.”6 And, almost a century and a half ago, Macaulay observed that the guilt of Titus Oates could not justify his conviction by improper methods: “That Oates was a bad man is not a sufficient excuse; for the guilty are almost always the first to suffer those hardships which are afterwards used as precedents against the innocent.”7

The Court’s holding focuses on the period after a suspect has been taken into custody and before he has been charged with an offense. The core of the Court’s holding is that police interference with an attorney’s access to her client during that period is not unconstitutional. The Court reasons that a State has a compelling interest, not simply in custodial interrogation, but in lawyer-free, incommunicado custodial interrogation. Such incommunicado interrogation is so important that a lawyer may be given false information that prevents her presence and representation; it is so important that police may refuse to inform a suspect of his attorney’s *438communications and immediate availability.8 This conclusion flies in the face of this Court’s repeated expressions of deep concern about incommunicado questioning.9 Until *439today, incommunicado questioning has been viewed with the strictest scrutiny by this Court; today, incommunicado questioning is embraced as a societal goal of the highest order that justifies police deception of the shabbiest kind.

It is not only the Court’s ultimate conclusion that is deeply disturbing; it is also its manner of reaching that conclusion. The Court completely rejects an entire body of law on the subject — the many carefully reasoned state decisions that have come to precisely the opposite conclusion.10 The Court *440similarly dismisses the fact that the police deception which it sanctions quite clearly violates the American Bar Association’s Standards for Criminal Justice11 — Standards which *441The Chief Justice has described as “the single most comprehensive and probably the most monumental undertaking in the field of criminal justice ever attempted by the American legal profession in our national history,”12 and which this Court frequently finds helpful.13 And, of course, the Court dismisses the fact that the American Bar Association has emphatically endorsed the prevailing state-court position and expressed its serious concern about the effect that a contrary view — a view, such as the Court’s, that exalts incommunicado interrogation, sanctions police deception, and demeans the right to consult with an attorney — will have in police stations and courtrooms throughout this Nation.14 Of greatest importance, the Court misapprehends or rejects the central principles that have, for several decades, animated this Court’s decisions concerning incommunicado interrogation.15

Police interference with communications between an attorney and his client is a recurrent problem. The factual variations in the many state-court opinions condemning this interference as a violation of the Federal Constitution suggest the *442variety of contexts in which the problem emerges. In Oklahoma, police led a lawyer to several different locations while they interrogated the suspect;16 in Oregon, police moved a suspect to a new location when they learned that his lawyer was on his way;17 in Illinois, authorities failed to tell a suspect that his lawyer had arrived at the jail and asked to see him;18 in Massachusetts, police did not tell suspects that their lawyers were at or near the police station.19 In all these cases, the police not only failed to inform the suspect, but also misled the attorneys. The scenarios vary, but the core problem of police interference remains. “Its recurrence suggests that it has roots in some condition fundamental and general to our criminal system.” Watts v. Indiana, 338 U. S. 49, 57 (1949) (Jackson, J., concurring in result).

The near-consensus of state courts and the legal profession’s Standards about this recurrent problem lends powerful support to the conclusion that police may not interfere with communications between an attorney and the client whom they are questioning. Indeed, at least two opinions from this Court seemed to express precisely that view.20 The Court today flatly rejects that widely held view and responds to this recurrent problem by adopting the most restrictive interpretation of the federal constitutional restraints on police *443deception, misinformation, and interference in attorney-client communications.

The exact reach of the Court’s opinion is not entirely clear because, on the one hand, it indicates that more egregious forms of police deception might violate the Constitution, ante, at 432, while, on the other hand, it endeavors to make its disposition of this case palatable by making findings of fact concerning the voluntariness of Burbine’s confessions that the trial judge who heard the evidence declined to make.21 Before addressing the legal issues, it therefore seems appropriate to make certain additional comments about what the record discloses concerning the incriminating statements made by Burbine during the 21-hour period that he was detained by the Cranston and Providence police on June 29 and June 30, 1977.

I

As the majority points out, with respect to attorney Munson’s telephone call, the Rhode Island Supreme Court’s summary of factual findings provides the common ground for analysis:

“At approximately 8:15 [on June 29, 1977], Ms. Mun-son called the Cranston police station and asked that her call be transferred to the detective division. A male voice responded with the word ‘Detectives.’ Ms. Mun-son identified herself and asked if Brian Burbine was being held; the person responded affirmatively. Ms. Munson explained to the person that Burbine was represented by attorney Casparian who was not available; she further stated that she would act as Burbine’s legal counsel in the event that the police intended to place him in a lineup or question him. The unidentified person told Ms. Munson that the police would not be questioning Burbine or putting him in a lineup and that they were *444through with him for the night. Ms. Munson was not informed that the Providence police were at the Cran-ston police station or that Burbine was a suspect in Mary’s murder. The trial justice found as a fact that Ms. Munson did make the call, but further found that there was no collusion or conspiracy on the part of the police ‘to secrete [Burbine] from his attorney . . . State v. Burbine, 451 A. 2d 22, 23-24 (1982).22

Although this paragraph accurately describes attorney Munson’s 8:15 call, the significance of the false response to her inquiry is best understood in the context of the events that were then proceeding in the police station. The difficulty in reconstructing some of those events illustrates the need for strict presumptions regarding the consequences of custodial interrogation — a need this Court has repeatedly recognized.23

*445On June 27, 1977, an unidentified person advised Detective Ferranti that a man known as “Butch,” who lived at 306 New York Avenue in Providence, was responsible for the death of Mary Jo Hickey. The record does not explain why Ferranti, who was a member of the Cranston Police Force, was informed about a crime that occurred in Providence.

At about 3 p.m. on June 29, 1977, Cranston police officers apprehended respondent Burbine and two other men (DiOrio and Sparks) in “a burned out building in the Cranston area.” S. H. 6, 180. The three men were taken to the Cranston police station, charged with “breaking and entering,” and placed in separate rooms. After noticing that DiOrio and Burbine lived at 306 New York Avenue in Providence, Detective Ferranti talked to DiOrio and was told that Burbine was the only “Butch” at that address. Id., at 146-147.

At approximately 4:30, Ferranti “went in the room where Burbine was” and asked him “if there was anybody that he knew by the name of Butch on the street, and he said he was the only Butch.” Id., at 148.24 After the brief questioning about the identity of “Butch,” Detective Ferranti left Burbine in the interrogation room — where he remained until about 9 p.m.25 — and interrogated DiOrio and Sparks. They both “made damaging statements relative to Burbine being involved in the murder in Providence”; Ferranti therefore “immediately contacted Providence Police.” Id., at 149-150. The Providence officers — Captain Wilson (the Chief of Detectives), Lieutenant Gannon, and Detective Trafford — responded promptly, and arrived at the Cranston station be*446tween 6 and 7 p.m. Lieutenant Gannon testified that, as he drove to the Cranston police station, he knew that he might not be able to question Burbine “[i]f for some reason he didn’t want to give me a statement, if for some reason he chose to get an attorney and the attorney informed us that he didn’t want him to give a statement.” Trial Tr. 407.

After arriving at the station, the three Providence officers, as well as Ferranti and a second Cranston officer (Lieutenant Ricard), either remained in the large central room in the basement of the Cranston police station, or participated in the questioning of DiOrio and Sparks in interrogation rooms adjacent to that large central room.

It was at this point — with Burbine alone in another adjacent room, with Providence police on hand, with police from two Departments questioning Sparks and DiOrio about Burbine’s involvement in the Hickey homicide — that attorney Munson telephoned. Her call arrived at 8:15; she asked for “Detectives,” and was told that the police “would not be questioning Burbine” and that they were “through” with him for the night. These statements were false. Moreover, she was not told that Burbine would be questioned about a homicide rather than the breaking-and-entering charge on which he had been arrested, and she was not told that Providence police were at the Cranston police station preparing to question Burbine about a Providence crime.

At about 9, some 45 minutes after Munson received the assurance that the police were “through” with Burbine, the officers completed their questioning of DiOrio and Sparks and were prepared to question Burbine. There is no dispute about the fact that Burbine was brought into the central room at about 9, that all five police officers were then present, and that Burbine appeared somewhat upset and professed that he “‘didn’t do anything wrong.’” S. H. 21. Detective Ferranti testified that this statement was in response to questions from the Providence police about the Hickey *447homicide;26 Lieutenant Gannon of the Providence police testified that the statement was about the Hickey homicide, but that Providence police did not question Burbine and that they merely saw Burbine being escorted by Ferranti.27 Burbine was not told that attorney Munson had called and had asked about him; nor was he told that Munson had been informed that the police were through with him for the night. After his protestations, Burbine was taken into another interrogation room.

Detective Ferranti then went into that room and, according to the testimony of the Providence officers, spent either “ten minutes” or from “five to ten minutes” alone with Burbine.28 The record does not tell us whether he told Burbine that Sparks and DiOrio had just given statements implicating him in the Hickey homicide. Nor does it resolve the question whether Burbine’s decision to confess was made before his session with Ferranti or as a result of that session. The Court evidently makes the former assumption, for it asserts that Burbine “initiated” this encounter. Ante, at 421-422. However, the state courts made no finding about this *448“initiation” by Burbine. Detective Ferranti testified that Burbine banged and kicked on the door, S. H. 153-154; Lieutenant Gannon testified that he “believed” there was a knocking or some communication from Burbine, id., at 22, but he was “not sure.” Id., at 66.29 None of the other officers, who were apparently in the large room adjacent to Burbine’s, corroborated this testimony by mentioning any “banging,” “kicking,” or other noise from Burbine’s direction. In all events, some minutes later, Detective Ferranti came back out of the room and indicated that Burbine wanted to talk.

Lieutenant Gannon and Detective Tráfford of the Providence police accompanied Detective Ferranti “back into the room.” During the period between 9:30 and 10:20 p.m., they administered Miranda warnings and typed out a four-page statement which Burbine signed, waiving his constitutional rights, acknowledging his responsibility for the death of Hickey, and reciting his version of that event. Ferranti alternately testified that Burbine was “coherent” and “incoherent” at the time of this questioning. Id., at 157-158; Trial Tr. 198, 208-209. Apparently for the first time since his arrival at the station in the afternoon, the police then brought Burbine some food. S. H. 160, Trial Tr. 205.

After obtaining Burbine’s signature on the first written statement at 10:20 p.m., the police were still not “through” with Burbine. Burbine’s first statement included no mention of the clothes that he had been wearing, or of a glass that was found with Hickey’s purse a few blocks from the homicide. Soon after the completion of the first statement, and after the Providence and Cranston officers had discussed the first statement and expressed pleasure with their success,30 *449Gannon, Trafford, and Ferranti again questioned Burbine. They ascertained that he was wearing his “red toke” and “black windbreaker” at the time, and that Hickey had left the bar with a glass in hand.31 At 11:20 p.m., Burbine signed the second statement.

The following morning, the officers obtained a warrant, conducted a search of Burbine’s residence, and seized the clothing that he had described in the second statement. In the meantime, Burbine was arraigned in Cranston court on the charge for which he had been arrested. Still without counsel, Burbine pleaded guilty to malicious damage. After the Cranston proceeding, Providence officers instantly arr rested him for the Hickey homicide. Trial Tr. 501. Burbine was taken to the Providence police station, where he executed a third waiver of rights and identified the coat and jacket that the officers had seized. Shortly after noon, Major Leyden called the Public Defender’s Office and requested counsel for Burbine because he would be placed in a lineup. Id., at 423.

Thus, although there are a number of ambiguities in the record, the state-court findings established (1) that attorney Munson made her call at about 8:15 p.m.; (2) that she was given false information; (3) that Burbine was not told of her *450call; and (4) that he was thereafter given the Miranda warnings, waived his rights, and signed three incriminating statements without receiving any advice from an attorney. The remainder of the record underscores two points. The first is the context of the call — a context in which two Police Departments were on the verge of resolving a highly publicized, hauntingly brutal homicide and in which, as Lieutenant Gannon testified, the police were aware that counsel’s advice to remain silent might be an obstacle to obtaining a confession. The second is the extent of the uncertainty about the events that motivated Burbine’s decision to waive his rights. The lawyer-free privacy of the interrogation room, so exalted by the majority, provides great difficulties in determining what actually transpired. It is not simply the ambiguity that is troublesome; if so, the problem would be not unlike other difficult evidentiary problems. Rather, the particularly troublesome aspect is that the ambiguity arises in the very situation — incommunicado interrogation — for which this Court has developed strict presumptions and for which this Court has, in the past, imposed the heaviest burden of justification on the goverment. It is in this context, and the larger context of our accusatorial system, that the deceptive conduct of the police must be evaluated.

l — l 1 — 1

Well-settled principles of law lead inexorably to the conclusion that the failure to inform Burbine of the call from his attorney makes the subsequent waiver of his constitutional rights invalid. Analysis should begin with an acknowledgment that the burden of proving the validity of a waiver of constitutional rights is always on the government,32 When *451such a waiver occurs in a custodial setting, that burden is an especially heavy one because custodial interrogation is inherently coercive,33 because disinterested witnesses are seldom available to describe what actually happened,34 and because history has taught us that the danger of overreaching during incommunicado interrogation is so real.35

In applying this heavy presumption against the validity of waivers, this Court has sometimes relied on a case-by-case totality of the circumstances analysis.36 We have found, however, that some custodial interrogation situations require strict presumptions against the validity of a waiver. Miranda established that a waiver is not valid in the absence of certain warnings. Edwards v. Arizona, 451 U. S. 477 (1981), similarly established that a waiver is not valid if police *452initiate questioning after the defendant has invoked his right to counsel. In these circumstances, the waiver is invalid as a matter of law even if the evidence overwhelmingly establishes, as a matter of fact, that “a suspect’s decision not to rely on his rights was uncoerced, that he at all times knew that he could stand mute and request a lawyer, and that he was aware of the State’s intention to use his statements to secure a conviction,” see ante, at 422. In light of our decision in Edwards, the Court is simply wrong in stating that “the analysis is complete and the waiver is valid as a matter of law” when these facts have been established. Ante, at 422-423.37 Like the failure to give warnings and like police initiation of interrogation after a request for counsel, police deception of a suspect through omission of information regarding attorney communications greatly exacerbates the inherent problems of incommunicado interrogation and requires a clear principle to safeguard the presumption against the waiver of constitutional rights. As in those situations, the police deception should render a subsequent waiver invalid.

Indeed, as Miranda itself makes clear, proof that the required warnings have been given is a necessary, but by no means sufficient, condition for establishing a valid waiver. As the Court plainly stated in Miranda, “any evidence that the accused was threatened, tricked, or cajoled into a waiver will, of course, show that the defendant did not voluntarily waive his privilege. The requirement of warnings and waiver of rights is a fundamental with respect to the Fifth *453Amendment privilege and not simply a preliminary ritual to existing methods of interrogation.” 384 U. S., at 476.

In this case it would be perfectly clear that Burbine’s waiver was invalid if, for example, Detective Ferranti had “threatened, tricked, or cajoled” Burbine in their private preconfession meeting — perhaps by misdescribing the statements obtained from DiOrio and Sparks — even though, under the Court’s truncated analysis of the issue, Burbine fully understood his rights. For Miranda clearly condemns threats or trickery that cause a suspect to make an unwise waiver of his rights even though he fully understands those rights. In my opinion there can be no constitutional distinction — as the Court appears to draw, ante, at 423-424 — between a deceptive misstatement and the concealment by the police of the critical fact that an attorney retained by the accused or his family has offered assistance, either by telephone or in person.38

Thus, the Court’s truncated analysis, which relies in part on a distinction between deception accomplished by means of an omission of a critically important fact and deception by means of a misleading statement, is simply untenable. If, as the Court asserts, “the analysis is at an end” as soon as the suspect is provided with enough information to have the capacity to understand and exercise his rights, I see no reason why the police should not be permitted to make the same kind of misstatements to the suspect that they are apparently allowed to make to his lawyer. Miranda, however, clearly *454establishes that both kinds of deception vitiate the suspect’s waiver of his right to counsel.39

As the Court notes, the question is whether the deceptive police conduct “deprives a defendant of knowledge essential to his ability to understand the nature of his rights and the consequences of abandoning them.” Ante, at 424. This question has been resoundingly answered time and time again by the state courts that, with rare exceptions,40 have correctly understood the meaning of the Miranda opinion.41 The ma*455jority’s blithe assertion of “no doubt” about the outcome of this case, ante, at 421, simply ignores the prevailing view of the state courts that have considered this issue. Particularly in an opinion that relies on a desire to avoid “a federal intrusion into the criminal processes of the States,” ante, at 434, one would expect at least some indication why, in the majority’s view, so many state courts have been so profoundly wrong on this precise issue. Unlike the majority, the state courts have realized that attorney communication to the police *456about the client is an event that has a direct “bearing” on the knowing and intelligent waiver of constitutional rights. As the Oregon Supreme Court has explained: “To pass up an abstract offer to call some unknown lawyer is very different from refusing to talk with an identified attorney actually available to provide at least initial assistance and advice, whatever might be arranged in the long run. A suspect indifferent to the first offer may well react quite differently to the second.” State v. Haynes, 288 Ore. 59, 72, 602 P. 2d 272, 278 (1979), cert. denied, 446 U. S. 945 (1980).42

In short, settled principles about construing waivers of constitutional rights and about the need for strict presumptions in custodial interrogations, as well as a plain reading of the Miranda opinion itself, overwhelmingly support the conclusion reached by almost every state court that has considered the matter — a suspect’s waiver of his right to counsel is invalid if police refuse to inform the suspect of his counsel’s communications.

) — f ) — I

The Court makes the alternative argument that requiring police to inform a suspect of his attorney’s communications to *457and about him is not required because it would upset the careful “balance” of Miranda. Despite its earlier notion that the attorney’s call is an “outside event” that has “no bearing” on a knowing and intelligent waiver, the majority does acknowledge that information of attorney Munson’s call “would have been useful to respondent” and “might have affected his decision to confess.” Ante, at 422.43 Thus, a rule requiring the police to inform a suspect of an attorney’s call would have two predictable effects. It would serve “Miranda’s goal of dispelling the compulsion inherent in custodial interrogation,” ante, at 425, and it would disserve the goal of custodial interrogation because it would result in fewer confessions. By a process of balancing these two concerns, the Court finds the benefit to the individual outweighed by the “substantial cost to society’s legitimate and substantial interest in securing admissions of guilt.” Ante, at 427.

The Court’s balancing approach is profoundly misguided. The cost of suppressing evidence of guilt will always make the value of a procedural safeguard appear “minimal,” “marginal,” or “incremental.” Indeed, the value of any trial at all seems like a “procedural technicality” when balanced against the interest in administering prompt justice to a murderer or a rapist caught redhanded. The individual interest in procedural safeguards that minimize the risk of error is easily discounted when the fact of guilt appears certain beyond doubt.

What is the cost of requiring the police to inform a suspect of his attorney’s call? It would decrease the likelihood that custodial interrogation will enable the police to obtain a confession. This is certainly a real cost, but it is the same cost that this Court has repeatedly found necessary to preserve *458the character of our free society and our rejection of an inquisitorial system. Three examples illustrate the point.

In Escobedo v. Illinois, 378 U. S. 478 (1964), we excluded a confession by a defendant who had not been permitted to consult with his lawyer, and whose lawyer had not been permitted to see him. We emphasized the “lesson of history” that our system of justice is not founded on a fear that a suspect will exercise his rights. “If the exercise of constitutional rights will thwart the effectiveness of a system of law enforcement, then there is something very wrong with that system.” Id., at 490. In Miranda v. Arizona, 384 U. S. 436 (1966), we similarly stressed this character of our system, despite its “cost,” by unequivocally holding that an individual has an absolute right to refuse to respond to police interrogation and to have the assistance of counsel during any questioning.44 Thus, as a matter of law, the assumed right of the police to interrogate a suspect is no right at all; at best, it is a mere privilege terminable at the will of the suspect. And, more recently in Dunaway v. New York, 442 U. S. 200 (1979), the Court corrected the long-held but mistaken view of the police that they have some sort of right to take any sus*459pect into custody for the purpose of questioning him even though they may not have probable cause to arrest.45

Just as the “cost” does not justify taking a suspect into custody or interrogating him without giving him warnings simply because police desire to question him, so too the “cost” does not justify permitting police to withhold from a suspect knowledge of an attorney’s communication, even though that communication would have an unquestionable effect on the suspect’s exercise of his rights. The “cost” that, concerns the Court amounts to nothing more than an acknowledgment that the law enforcement interest in obtaining convictions suffers whenever a suspect exercises the rights that are afforded by our system of criminal justice. In other words, it is the fear that an individual may exercise his rights that tips the scales of justice for the Court today. The principle that ours is an accusatorial, not an inquisitorial, system, however, has repeatedly led the Court to reject that fear as a valid reason for inhibiting the invocation of rights.

If the Court’s cost-benefit analysis were sound, it would justify a repudiation of the right to a warning about counsel itself. There is only a difference in degree between a presumption that advice about the immediate availability of a lawyer would not affect the voluntariness of a decision to confess, and a presumption that every citizen knows that he has a right to remain silent and therefore no warnings of any kind are needed. In either case, the withholding of information serves precisely the same law enforcement interests. And in both cases, the cost can be described as nothing more than *460an incremental increase in the risk that an individual will make an unintelligent waiver of his rights.

In cases like Escobedo, Miranda, and Dunaway, the Court has viewed the balance from a much broader perspective. In all these cases — indeed, whenever the distinction between an inquisitorial and an accusatorial system of justice is implicated — the law enforcement interest served by incommunicado interrogation has been weighed against the interest in individual liberty that is threatened by such practices. The balance has never been struck by an evaluation of empirical data of the kind submitted to legislative decisionmakers —indeed, the Court relies on no such data today. Rather, the Court has evaluated the quality of the conflicting rights and interests. In the past, that kind of balancing process has led to the conclusion that the police have no right to compel an individual to respond to custodial interrogation, and that the interest in liberty that is threatened by incommunicado interrogation is so precious that special procedures must be followed to protect it. The Court’s contrary conclusion today can only be explained by its failure to appreciate the value of the liberty that an accusatorial system seeks to protect.

> I — I

The Court also argues that a rule requiring the police to inform a suspect of an attorney’s efforts to reach him would have an additional cost: it would undermine the “clarity” of the rule of the Miranda case. Ante, at 425-426. This argument is not supported by any reference to the experience in the States that have adopted such a rule. The Court merely professes concern about its ability to answer three quite simple questions.46

*461Moreover, the Court’s evaluation of the interest in “clarity” is rather one-sided. For a police officer with a printed card containing the exact text he is supposed to recite, perhaps the rule is clear. But the interest in clarity that the Miranda decision was intended to serve is not merely for the benefit of the police. Rather, the decision was also, and primarily, intended to provide adequate guidance to the person in custody who is being asked to waive the protections afforded by the Constitution.47 Inevitably, the Miranda decision also serves the judicial interest in clarifying the inquiry *462into what actually transpired during a custodial interrogation.48 Under the Court’s conception of the interest in clarity, however, the police would presumably prevail whenever they could convince the trier of fact that a required ritual was performed before the confession was obtained.

V

At the time attorney Munson made her call to the Cranston police station, she was acting as Burbine’s attorney. Under ordinary principles of agency law the deliberate deception of Munson was tantamount to deliberate deception of her client.49 If an attorney makes a mistake in the course of her representation of her client, the client must accept the consequences of that mistake.50 It is equally clear that when an attorney makes an inquiry on behalf of her client, the client is entitled to a truthful answer. Surely the client must have the same remedy for a false representation to his lawyer that he would have if he were acting pro se and had propounded the question himself.

The majority brushes aside the police deception involved in the misinformation of attorney Munson. It is irrelevant to the Fifth Amendment analysis, concludes the majority, because that right is personal; it is irrelevant to the Sixth *463Amendment analysis, continues the majority, because the Sixth Amendment does not apply until formal adversary proceedings have begun.

In my view, as a matter of law, the police deception of Munson was tantamount to deception of Burbine himself. It constituted a violation of Burbine’s right to have an attorney present during the questioning that began shortly thereafter. The existence of that right is undisputed.51 Whether the source of that right is the Sixth Amendment, the Fifth Amendment, or a combination of the two is of no special importance, for I do not understand the Court to deny the existence of the right.

The pertinent question is whether police deception of the attorney is utterly irrelevant to that right. In my judgment, it blinks at reality to suggest that misinformation which prevented the presence of an attorney has no bearing on the protection and effectuation of the right to counsel in custodial interrogation. The majority parses the role of attorney and suspect so narrowly that the deception of the attorney is of no *464constitutional significance. In other contexts, however, the Court does not hesitate to recognize an identity between the interest of attorney and accused.52 The character of the attorney-client relationship requires rejection of the Court’s notion that the attorney is some entirely distinct, completely severable entity and that deception of the attorney is irrelevant to the right of counsel in custodial interrogation.53

*465The possible reach of the Court’s opinion is stunning. For the majority seems to suggest that police may deny counsel all access to a client who is being held. At least since Escobedo v. Illinois, it has been widely accepted that police may not simply deny attorneys access to their clients who are in custody. This view has survived the recasting of Escobedo from a Sixth Amendment to a Fifth Amendment case that the majority finds so critically important. That this prevailing view is shared by the police can be seen in the state-court opinions detailing various forms of police deception of attorneys.54 For, if there were no obligation to give attorneys access, there would be no need to take elaborate steps to avoid access, such as shuttling the suspect to a different location,55 or taking the lawyer to different locations;56 police could simply refuse to allow the attorneys to see the suspects. But the law enforcement profession has apparently believed, quite rightly in my view, that denying lawyers access to their clients is impermissible. The Court today seems to assume that this view was error — that, from the federal constitutional perspective, the lawyer’s access is, as a question from the Court put it in oral argument, merely “a matter of prosecutorial grace.” Tr. of Oral Arg. 32. Certainly, nothing in the Court’s Fifth and Sixth Amendment analysis acknowledges that there is any federal constitutional bar to an absolute denial of lawyer access to a suspect who is in police custody.

In sharp contrast to the majority, I firmly believe that the right to counsel at custodial interrogation is infringed by police treatment of an attorney that prevents or impedes the attorney’s representation of the suspect at that interrogation.

*466VI

The Court devotes precisely five sentences to its conclusion that the police interference in the attorney’s representation of Burbine did not violate the Due Process Clause. In the majority’s view, the due process analysis is a simple “shock the conscience” test. Finding its conscience troubled,57 but not shocked, the majority rejects the due process challenge.

In a variety of circumstances, however, the Court has given a more thoughtful consideration to the requirements of due process. For instance, we have concluded that use of a suspect’s post-Miranda warnings silence against him violates the due process requirement of fundamental fairness because such use breaches an implicit promise that “silence will carry no penalty.”58 Similarly, we have concluded that “the suppression by the prosecution of evidence favorable to an accused upon request violates due process where the evidence is material either to guilt or to punishment.”59 We have also concluded that vindictive prosecution violates due process;60 so too does vindictive sentencing.61 Indeed, we have emphasized that analysis of the “voluntariness” of a confession is frequently a “convenient shorthand” for reviewing objectionable police methods under the rubric of the due process requirement of fundamental fairness.62 What emerges from *467these cases is not the majority’s simple “shock the conscience” test, but the principle that due process requires fairness, integrity, and honor in the operation of the criminal justice system, and in its treatment of the citizen’s cardinal constitutional protections.

In my judgment, police interference in the attorney-client relationship is the type of governmental misconduct on a matter of central importance to the administration of justice that the Due Process Clause prohibits. Just as the police cannot impliedly promise a suspect that his silence will not be used against him and then proceed to break that promise, so too police cannot tell a suspect’s attorney that they will not question the suspect and then proceed to question him. Just as the government cannot conceal from a suspect material and exculpatory evidence, so too the government cannot conceal from a suspect the material fact of his attorney’s communication.

*468Police interference with communications between an attorney and his client violates the due process requirement of fundamental fairness. Burbine’s attorney was given completely false information about the lack of questioning; moreover, she was not told that her client would be questioned regarding a murder charge about which she was unaware. Burbine, in turn, was not told that his attorney had phoned and that she had been informed that he would not be questioned. Quite simply, the Rhode Island police effectively drove a wedge between an attorney and a suspect through misinformation and omissions.

The majority does not “question that on facts more egregious than those presented here police deception might rise to a level of a due process violation.” Ante, at 432. In my view, the police deception disclosed by this record plainly does rise to that level.

VII

This case turns on a proper appraisal of the role of the lawyer in our society. If a lawyer is seen as a nettlesome obstacle to the pursuit of wrongdoers — as in an inquisitorial society — then the Court’s decision today makes a good deal of sense. If a lawyer is seen as an aid to the understanding and protection of constitutional rights — as in an accusatorial society — then today’s decision makes no sense at all.

Like the conduct of the police in the Cranston station on the evening of June 29, 1977, the Court’s opinion today serves the goal of insuring that the perpetrator of a vile crime is punished. Like the police on that June night as well, however, the Court has trampled on well-established legal principles and flouted the spirit of our accusatorial system of justice.

I respectfully dissent.

Justice Kennedy

delivered the opinion of the Court.

The United States Court of Appeals for the Sixth Circuit, in a habeas corpus proceeding challenging a Michigan conviction for first-degree murder and certain other offenses, ruled that there had been two separate constitutional errors in the trial that led to the jury's guilty verdict. First, the Court *374of Appeals determined that a statement by the accused, relied on at trial by the prosecution, had been elicited in violation of Miranda v. Arizona, 384 U. S. 436 (1966). Second, it found that failure to ask for an instruction relating to testimony from an accomplice was ineffective assistance by defense counsel. See Strickland v. Washington, 466 U. S. 668 (1984). Both of these contentions had been rejected in Michigan courts and in the habeas corpus proceedings before the United States District Court. Certiorari was granted to review the decision by the Court of Appeals on both points. The warden of a Michigan correctional facility is the petitioner here, and Van Chester Thompkins, who was convicted, is the respondent.

I

A

On January 10, 2000, a shooting occurred outside a mall in Southfield, Michigan. Among the victims was Samuel Morris, who died from multiple gunshot wounds. The other victim, Frederick France, recovered from his injuries and later testified. Thompkins, who was a suspect, fled. About one year later he was found in Ohio and arrested there.

Two Southfield police officers traveled to Ohio to interrogate Thompkins, then awaiting transfer to Michigan. The interrogation began around 1:30 p.m. and lasted about three hours. The interrogation was conducted in a room that was 8 by 10 feet, and Thompkins sat in a chair that resembled a school desk (it had an arm on it that swings around to provide a surface to write on). App. 144a-145a. At the beginning of the interrogation, one of the officers, Detective Hel-gert, presented Thompkins with a form derived from the Miranda rule. It stated:

"NOTIFICATION OF CONSTITUTIONAL RIGHTS AND STATEMENT

"1. You have the right to remain silent.

*375“2. Anything you say can and will be used against you in a court of law.

“3. You have a right to talk to a lawyer before answering any questions and you have the right to have a lawyer present with you while you are answering any questions.

“4. If you cannot afford to hire a lawyer, one will be appointed to represent you before any questioning, if you wish one.

“5. You have the right to decide at any time before or during questioning to use your right to remain silent and your right to talk with a lawyer while you are being questioned.” Brief for Petitioner 60 (some capitalization omitted).

Helgert asked Thompkins to read the fifth warning out loud. App. 8a. Thompkins complied. Helgert later said this was to ensure that Thompkins could read, and Helgert concluded that Thompkins understood English. Id., at 9a. Helgert then read the other four Miranda warnings out loud and asked Thompkins to sign the form to demonstrate that he understood his rights. App. 8a-9a. Thompkins declined to sign the form. The record contains conflicting evidence about whether Thompkins then verbally confirmed that he understood the rights listed on the form. Compare id., at 9a (at a suppression hearing, Helgert testified that Thompkins verbally confirmed that he understood his rights), with id., at 148a (at trial, Helgert stated, “I don’t know that I orally asked him” whether Thompkins understood his rights).

Officers began an interrogation. At no point during the interrogation did Thompkins say that he wanted to remain silent, that he did not want to talk with the police, or that he wanted an attorney. Id., at 10a. Thompkins was “[ljargely” silent during the interrogation, which lasted about three hours. Id., at 19a. He did give a few limited verbal responses, however, such as “yeah,” “no,” or “I don’t know.” And on occasion he communicated by nodding his *376head. Id., at 23a. Thompkins also said that he “didn’t want a peppermint” that was offered to him by the police and that the chair he was “sitting in was hard.” Id., at 152a.

About 2 hours and 45 minutes into the interrogation, Hel-gert asked Thompkins, “Do you believe in God?” Id., at 11a, 153a. Thompkins made eye contact with Helgert and said “Yes,” as his eyes “well[ed] up with tears.” Id., at 11a. Helgert asked, “Do you pray to God?” Thompkins said “Yes.” Id., at 11a, 153a. Helgert asked, “Do you pray to God to forgive you for shooting that boy down?” Id., at 153a. Thompkins answered “Yes” and looked away. Ibid. Thompkins refused to make a written confession, and the interrogation ended about 15 minutes later. Id., at 11a.

Thompkins was charged with first-degree murder, assault with intent to commit murder, and certain firearms-related offenses. He moved to suppress the statements made during the interrogation. He argued that he had invoked his Fifth Amendment right to remain silent, requiring police to end the interrogation at once, see Michigan v. Mosley, 423 U. S. 96, 103 (1975) (citing Miranda, 384 U. S., at 474), that he had not waived his right to remain silent, and that his inculpatory statements were involuntary. The trial court denied the motion.

At trial, the prosecution’s theory was that Thompkins shot the victims from the passenger seat of a van driven by Eric Purifoy. Purifoy testified that he had been driving the van and that Thompkins was in the passenger seat while another man, one Myzell Woodward, was in the back. The defense strategy was to pin the blame on Purifoy. Purifoy testified he did not see who fired the weapon because the van was stopped and he was bending over near the floor when shots were fired. Purifoy explained that, just after the shooting, Thompkins, holding a pistol, told Purifoy, “What the hell you doing? Pull off.” Purifoy then drove away from the scene. App. 170a.

*377So that the Thompkins jury could assess Purifoy’s credibility and knowledge, the prosecution elicited testimony from Purifoy that he had been tried earlier for the shooting under an aiding-and-abetting theory. Purifoy and Detective Hel-gert testified that a jury acquitted him of the murder and assault charges, convicted him of carrying a concealed weapon in a motor vehicle, and hung on two other firearms offenses to which he later pleaded guilty. At Purifoy’s trial, the prosecution had argued that Purifoy was the driver and Thompkins was the shooter. This was consistent with the prosecution’s argument at Thompkins’ trial.

After Purifoy’s trial had ended — but before Thompkins’ trial began — Purifoy sent Thompkins some letters. The letters expressed Purifoy’s disappointment that Thompkins’ family thought Purifoy was a “snitch” and a “rat.” Id., at 179a-180a. In one letter Purifoy offered to send a copy of his trial transcript to Thompkins as proof that Purifoy did not place the blame on Thompkins for the shooting. Id., at 180a. The letters also contained statements by Purifoy that claimed they were both innocent. Id., at 178a-179a. At Thompkins’ trial, the prosecution suggested that one of Purifoy’s letters appeared to give Thompkins a trial strategy. It was, the prosecution suggested, that Woodward shot the victims, allowing Purifoy and Thompkins to say they dropped to the floor when the shooting started. Id., at 187a-189a.

During closing arguments, the prosecution suggested that Purifoy lied when he testified that he did not see Thompkins shoot the victims:

“Did Eric Purifoy’s Jury make the right decision? I’m not here to judge that. You are not bound by what his Jury found. Take his testimony for what it was, [a] twisted attempt to help not just an acquaintance but his tight buddy.” Id., at 202a.

*378Defense counsel did not object. Defense counsel also did not ask for an instruction informing the jury that it could consider evidence of the outcome of Purifoy’s trial only to assess Purifoy’s credibility, not to establish Thompkins’ guilt.

The jury found Thompkins guilty on all counts. He was sentenced to life in prison without parole.

B

The trial court denied a motion for new trial filed by Thompkins’ appellate counsel. The trial court rejected the claim of ineffective assistance of trial counsel for failure to ask for a limiting instruction regarding the outcome of Puri-foy’s trial, reasoning that this did not prejudice Thompkins. Id., at 236a.

Thompkins appealed this ruling, along with the trial court’s refusal to suppress his pretrial statements under Miranda. The Michigan Court of Appeals rejected the Miranda claim, ruling that Thompkins had not invoked his right to remain silent and had waived it. It also rejected the ineffective-assistance-of-counsel claim, finding that Thompkins failed to show that evidence of Purifoy’s conviction for firearms offenses resulted in prejudice. People v. Thompkins, No. 242478, (Feb. 3, 2004), App. to Pet. for Cert. 74a-82a. The Michigan Supreme Court denied discretionary review. 471 Mich. 866, 683 N. W. 2d 676 (2004) (table).

Thompkins filed a petition for a writ of habeas corpus in the United States District Court for the Eastern District of Michigan. The District Court rejected Thompkins’ Miranda and ineffective-assistance claims. App. to Pet. for Cert. 39a-72a. It noted that, under the Antiterrorism and Effective Death Penalty Act of 1996 (AEDPA), a federal court cannot grant a petition for a writ of habeas corpus unless the state court’s adjudication of the merits was “contrary to, or involved an unreasonable application of, clearly established Federal law.” 28 U. S. C. § 2254(d)(1). The District Court reasoned that Thompkins did not invoke his right to remain silent and was not coerced into making statements *379during the interrogation. It held further that the Michigan Court of Appeals was not unreasonable in determining that Thompkins had waived his right to remain silent.

The United States Court of Appeals for the Sixth Circuit reversed, ruling for Thompkins on both his Miranda and ineffeetive-assistance-of-counsel claims. 547 F. 3d 572 (2008). The Court of Appeals ruled that the state court, in rejecting Thompkins’ Miranda claim, unreasonably applied clearly established federal law and based its decision on an unreasonable determination of the facts. See 28 U. S. C. § 2254(d). The Court of Appeals acknowledged that a waiver of the right to remain silent need not be express, as it can be “ ‘inferred from the actions and words of the person interrogated.’” 547 F. 3d, at 582 (quoting North Carolina v. Butler, 441 U. S. 369, 373 (1979)). The panel held, nevertheless, that the state court was unreasonable in finding an implied waiver in the circumstances here. The Court of Appeals found that the state eourt unreasonably determined the facts because “the evidence demonstrates that Thomp-kins was silent for two hours and forty-five minutes.” 547 F. 3d, at 586. According to the Court of Appeals, Thomp-kins’ “persistent silence for nearly three hours in response to questioning and repeated invitations to tell his side of the story offered a clear and unequivocal message to the officers: Thompkins did not wish to waive his rights.” Id., at 588.

The Court of Appeals next determined that the state court unreasonably applied clearly established federal law by rejecting Thompkins’ ineffective-assistance-of-eounsel claim based on counsel’s failure to ask for a limiting instruction regarding Purifoy’s acquittal. The Court of Appeals asserted that because Thompkins’ central strategy was to pin the blame on Purifoy, there was a reasonable probability that the result of Thompkins’ trial would have been different if there had been a limiting instruction regarding Purifoy’s acquittal.

We granted certiorari. 557 U. S. 965 (2009).

*380II

Under AEDPA, a federal court may not grant a habeas corpus application “with respect to any claim that was adjudicated on the merits in State court proceedings,” 28 U. S. C. § 2254(d), unless the state court’s decision “was contrary to, or involved an unreasonable application of, clearly established Federal law, as determined by the Supreme Court of the United States,” § 2254(d)(1), or “was based on an unreasonable determination of the facts in light of the evidence presented in the State court proceeding,” § 2254(d)(2). See Knowles v. Mirzayance, 556 U. S. 111, 114 (2009). The relevant state-court decision here is the Michigan Court of Appeals’ decision affirming Thompkins’ conviction and rejecting his Miranda and ineffeetive-assistance-of-counsel claims on the merits.

III

The Miranda Court formulated a warning that must be given to suspects before they can be subjected to custodial interrogation. The substance of the warning still must be given to suspects today. A suspect in custody must be advised as follows:

“He must be warned prior to any questioning that he has the right to remain silent, that anything he says can be used against him in a court of law, that he has the right to the presence of an attorney, and that if he cannot afford an attorney one will be appointed for him prior to any questioning if he so desires.” 384 U. S., at 479.

All concede that the warning given in this case was in full compliance with these requirements. The dispute centers on the response — or nonresponse — from the suspect.

A

Thompkins makes various arguments that his answers to questions from the detectives were inadmissible. He first *381contends that he “invoke[d] his privilege” to remain silent by not saying anything for a sufficient period of time, so the interrogation should have “cease[d]” before he made his in-culpatory statements. Id., at 474; see Mosley, 423 U. S., at 103 (police must “‘scrupulously hono[r]’” this “critical safeguard” when the accused invokes his or her “ ‘right to cut off questioning’” (quoting Miranda, supra, at 474, 479)).

This argument is unpersuasive. In the context of invoking the Miranda right to counsel, the Court in Davis v. United States, 512 U. S. 452, 459 (1994), held that a suspect must do so “unambiguously.” If an accused makes a statement concerning the right to counsel “that is ambiguous or equivocal” or makes no statement, the police are not required to end the interrogation, ibid., or ask questions to clarify whether the accused wants to invoke his or her Miranda rights, 512 U. S., at 461-462.

The Court has not yet stated whether an invocation of the right to remain silent can be ambiguous or equivocal, but there is no principled reason to adopt different standards for determining when an accused has invoked the Miranda right to remain silent and the Miranda right to counsel at issue in Davis. See, e. g., Solem v. Stumes, 465 U. S. 638, 648 (1984) (“[M]uch of the logic and language of [Mosley],” which discussed the Miranda right to remain silent, “could be applied to the invocation of the [Miranda right to counsel]”). Both protect the privilege against compulsory self-incrimination, Miranda, supra, at 467-473, by requiring an interrogation to cease when either right is invoked, Mosley, supra, at 103 (citing Miranda, supra, at 474); Fare v. Michael C., 442 U. S. 707, 719 (1979).

There is good reason to require an accused who wants to invoke his or her right to remain silent to do so unambiguously. A requirement of an unambiguous invocation of Miranda rights results in an objective inquiry that “avoid[s] difficulties of proof and . . . provide[s] guidance to officers” on how to proceed in the face of ambiguity. Davis, 512 U. S., *382at 458-459. If an ambiguous act, omission, or statement could require police to end the interrogation, police would be required to make difficult decisions about an accused’s unclear intent and face the consequence of suppression “if they guess wrong.” Id., at 461. Suppression of a voluntary confession in these circumstances would place a significant burden on society’s interest in prosecuting criminal activity. See id., at 459-461; Moran v. Burbine, 475 U. S. 412, 427 (1986). Treating an ambiguous or equivocal act, omission, or statement as an invocation of Miranda rights “might add marginally to Miranda’s goal of dispelling the compulsion inherent in custodial interrogation.” Burbine, 475 U. S., at 425. But “as Miranda holds, full comprehension of the rights to remain silent and request an attorney are sufficient to dispel whatever coercion is inherent in the interrogation process.” Id., at 427; see Davis, supra, at 460.

Thompkins did not say that he wanted to remain silent or that he did not want to talk with the police. Had he made either of these simple, unambiguous statements, he would have invoked his “'right to cut off questioning.’” Mosley, supra, at 103 (quoting Miranda, supra, at 474). Here he did neither, so he did not invoke his right to remain silent.

B

We next consider whether Thompkins waived his right to remain silent. Even absent the accused’s invocation of the right to remain silent, the accused’s statement during a custodial interrogation is inadmissible at trial unless the prosecution can establish that the accused “in fact knowingly and voluntarily waived [MirandaJ rights” when making the statement. Butler, 441 U. S., at 373. The waiver inquiry “has two distinct dimensions”: waiver must be “voluntary in the sense that it was the product of a free and deliberate choice rather than intimidation, coercion, or deception,” and “made with a full awareness of both the nature of the right *383being abandoned and the consequences of the decision to abandon it.” Burbine, supra, at 421.

Some language in Miranda could be read to indicate that waivers are difficult to establish absent an explicit written waiver or a formal, express oral statement. Miranda said “a valid waiver will not be presumed simply from the silence of the accused after warnings are given or simply from the fact that a confession was in fact eventually obtained.” 384 U. S., at 475; see id., at 470 (“No effective waiver . . . can be recognized unless specifically made after the [Miranda] warnings .. . have been given”). In addition, the Miranda Court stated that “a heavy burden rests on the government to demonstrate that the defendant knowingly and intelligently waived his privilege against self-incrimination and his right to retained or appointed counsel.” Id., at 475.

The course of decisions since Miranda, informed by the application of Miranda warnings in the whole course of law enforcement, demonstrates that waivers can be established even absent formal or express statements of waiver that would be expected in, say, a judicial hearing to determine if a guilty plea has been properly entered. Cf. Fed. Rule Crim. Proc. 11. The main purpose of Miranda is to ensure that an accused is advised of and understands the right to remain silent and the right to counsel. See Davis, supra, at 460; Burbine, supra, at 427. Thus, “[i]f anything, our subsequent cases have reduced the impact of the Miranda rule on legitimate law enforcement while reaffirming the decision’s core ruling that unwarned statements may not be used as evidence in the prosecution's case in chief.” Dickerson v. United States, 530 U. S. 428, 443-444 (2000).

One of the first cases to decide the meaning and import of Miranda with respect to the question of waiver was North Carolina v. Butler. The Butler Court, after discussing some of the problems created by the language in Miranda, established certain important propositions. Butler interpreted the Miranda language concerning the “heavy bur*384den” to show waiver, 384 U. S., at 475, in accord with usual principles of determining waiver, which can include waiver implied from all the circumstances. See Butler, supra, at 373, 376. And in a later case, the Court stated that this “heavy burden” is not more than the burden to establish waiver by a preponderance of the evidence. Colorado v. Connelly, 479 U. S. 157, 168 (1986).

The prosecution therefore does not need to show that a waiver of Miranda rights was express. An “implicit waiver” of the “right to remain silent” is sufficient to admit a suspect’s statement into evidence. Butler, supra, at 376. Butler made clear that a waiver of Miranda rights may be implied through “the defendant’s silence, coupled with an understanding of his rights and a course of conduct indicating waiver.” 441 U. S., at 373. The Court in Butler therefore “retreated” from the “language and tenor of the Miranda opinion,” which “suggested that the Court would require that a waiver ... be ‘specifically made.’” Connecticut v. Barrett, 479 U. S. 523, 531-532 (1987) (Brennan, J., concurring in judgment).

If the State establishes that a Miranda warning was given and the accused made, an uncoerced statement, this showing, standing alone, is insufficient to demonstrate “a valid waiver” of Miranda rights. Miranda, supra, at 475. The prosecution must make the additional showing that the accused understood these rights. See Colorado v. Spring, 479 U. S. 564, 573-575 (1987); Barrett, supra, at 530; Burbine, 475 U. S., at 421-422. Cf. Tague v. Louisiana, 444 U. S. 469, 469, 471 (1980) (per curiam) (no evidence that accused understood his Miranda rights); Carnley v. Cochran, 369 U. S. 506, 516 (1962) (government could not show that accused “understandingly” waived his right to counsel in light of “silent record”). Where the prosecution shows that a Miranda warning was given and that it was understood by the accused, an accused’s uncoerced statement establishes an implied waiver of the right to remain silent.

*385Although Miranda imposes on the police a rule that is both formalistic and practical when it prevents them from interrogating suspects without first providing them with a Miranda warning, see Burbine, 475 U. S., at 427, it does not impose a formalistic waiver procedure that a suspect must follow to relinquish those rights. As a general proposition, the law can presume that an individual who, with a full understanding of his or her rights, acts in a manner inconsistent with their exercise has made a deliberate choice to relinquish the protection those rights afford. See, e. g., Butler, supra, at 372-376; Connelly, supra, at 169-170 (“There is obviously no reason to require more in the way of a ‘volun-tariness’ inquiry in the Miranda waiver context than in the [due process] confession context”). The Court’s cases have recognized that a waiver of Miranda rights need only meet the standard of Johnson v. Zerbst, 304 U. S. 458, 464 (1938). See Butler, supra, at 374-375; Miranda, supra, at 475-476 (applying Zerbst standard of intentional relinquishment of a known right). As Butler recognized, 441 U. S., at 375-376, Miranda rights can therefore be waived through means less formal than a typical waiver on the record in a courtroom, cf. Fed. Rule Crim. Proc. 11, given the practical constraints and necessities of interrogation and the fact that Miranda’s main protection lies in advising defendants of their rights, see Davis, 512 U. S., at 460; Burbine, 475 U. S., at 427.

The record in this case shows that Thompkins waived his right to remain silent. There is no basis in this case to conclude that he did not understand his rights; and on these facts it follows that, he chose not to invoke or rely on those rights when he did speak. First, there is no contention that Thompkins did not understand his rights; and from this it follows that he knew what he gave up when he spoke. See id., at 421. There was more than enough evidence in the record to conclude that Thompkins understood his Miranda rights. Thompkins received a written copy of the Miranda warnings; Detective Helgert determined that Thompkins *386could read and understand English; and Thompkins was given time to read the warnings. Thompkins, furthermore, read aloud the fifth warning, which stated that “you have the right to deeide at any time before or during questioning to use your right to remain silent and your right to talk with a lawyer while you are being questioned.” Brief for Petitioner 60 (capitalization omitted). He was thus aware that his right to remain silent would not dissipate after a certain amount of time and that police would have to honor his right to be silent and his right to counsel during the whole course of interrogation. Those rights, the warning made clear, could be asserted at any time. Helgert, moreover, read the warnings aloud.

Second, Thompkins’ answer to Detective Helgert’s question about whether Thompkins prayed to God for forgiveness for shooting the victim is a “course of conduct indicating waiver” of the right to remain silent. Butler, supra, at 373. If Thompkins wanted to remain silent, he could have said nothing in response to Helgert’s questions, or he could have unambiguously invoked his Miranda rights and ended the interrogation. The fact that Thompkins made a statement about three hours after receiving a Miranda warning does not overcome the fact that he engaged in a course of conduct indicating waiver. Police are not required to rewarn suspects from time to time. Thompkins’ answer to Helgert’s question about praying to God for forgiveness for shooting the victim was sufficient to show a course of conduct indicating waiver. This is confirmed by the fact that before then Thompkins had given sporadic answers to questions throughout the interrogation.

Third, there is no evidence that Thompkins’ statement was coerced. See Burbine, supra, at 421. Thompkins does not claim that police threatened or injured him during the interrogation or that he was in any way fearful. The interrogation was conducted in a standard-sized room in the middle of the afternoon. It is true that apparently he was in a *387straight-backed chair for three hours, but there is no authority for the proposition that an interrogation of this length is inherently coercive. Indeed, even where interrogations of greater duration were held to be improper, they were accompanied, as this one was not, by other facts indicating coercion, such as an incapacitated and sedated suspect, sleep and food deprivation, and threats. Cf. Connelly, 479 U. S., at 163-164, n. 1. The fact that Helgert’s question referred to Thompkins’ religious beliefs also did not render Thompkins’ statement involuntary. “[T]he Fifth Amendment privilege is not concerned ‘with moral and psychological pressures to confess emanating from sources other than official coercion.’ ” Id., at 170 (quoting Oregon v. Elstoud, 470 U. S. 298, 305 (1985)). In these circumstances, Thompkins knowingly and voluntarily made a statement to police, so he waived his right to remain silent.

C

Thompkins next argues that, even if his answer to Detective Helgert could constitute a waiver of his right to remain silent, the police were not allowed to question him until they obtained a waiver first. Butler forecloses this argument. The Butler Court held that courts can infer a waiver of Miranda rights “from the actions and words of the person interrogated.” 441 U. S., at 373. This principle would be inconsistent with a rule that requires a waiver at the outset. The Butler Court thus rejected the rule proposed by the Butler dissent, which would have “requir[ed] the police to obtain an express waiver of [Miranda rights] before proceeding with interrogation.” Id., at 379 (Brennan, J., dissenting). This holding also makes sense given that “the primary protection afforded suspects subject[ed] to custodial interrogation is the Miranda warnings themselves.” Davis, supra, at 460. The Miranda rule and its requirements are met if a suspect receives adequate Miranda warnings, understands them, and has an opportunity to invoke the rights before giving any answers or admissions. Any waiver, express or implied, *388may be contradicted by an invocation at any time. If the right to counsel or the right to remain silent is invoked at any point during questioning, further interrogation must cease.

Interrogation provides the suspect with additional information that can put his or her decision to waive, or not to invoke, into perspective. As questioning commences and then continues, the suspect has the opportunity to consider the choices he or she faces and to make a more informed decision, either to insist on silence or to cooperate. When the suspect knows that Miranda rights can be invoked at any time, he or she has the opportunity to reassess his or her immediate and long-term interests. Cooperation with the police may result in more favorable treatment for the suspect; the apprehension of accomplices; the prevention of continuing injury and fear; beginning steps toward relief or solace for the victims; and the beginning of the suspect’s own return to the law and the social order it seeks to protect.

In order for an accused’s statement to be admissible at trial, police must have given the accused a Miranda warning. See Miranda, 384 U. S., at 471. If that condition is established, the court can proceed to consider whether there has been an express or implied waiver of Miranda rights. Id., at 476. In making its ruling on the admissibility of a statement made during custodial questioning, the trial court, of course, considers whether there is evidence to support the conclusion that, from the whole course of questioning, an express or implied waiver has been established. Thus, after giving a Miranda warning, police may interrogate a suspect who has neither invoked nor waived his or her Miranda rights. On these premises, it follows the police were not required to obtain a waiver of Thompkins’ Miranda rights before commencing the interrogation.

D

In sum, a suspeet who has received and understood the Miranda warnings, and has not invoked his Miranda rights, *389waives the right to remain silent by making an uncoerced statement to the police. Thompkins did not invoke his right to remain silent and stop the questioning. Understanding his rights in full, he waived his right to remain silent by making a voluntary statement to the police. The police, moreover, were not required to obtain a waiver of Thomp-kins’ right to remain silent before interrogating him. The state court’s decision rejecting Thompkins’ Miranda claim was thus correct under de novo review and therefore necessarily reasonable under the more deferential AEDPA standard of review, 28 U. S. C. § 2254(d). See Knowles, 556 U. S., at 123-124 (state court’s decision was correct under de novo review and not unreasonable under AEDPA).

IV

The second issue in this case is whether Thompkins’ counsel provided ineffective assistance by failing to request a limiting instruction regarding how the jury could consider the outcome of Purifoy’s trial. To establish ineffective assistance of counsel, a defendant “must show both deficient performance by counsel and prejudice.” Id., at 122 (citing Strickland, 466 U. S., at 687). To establish prejudice, a “defendant must show that there is a reasonable probability that, but for counsel’s unprofessional errors, the result of the proceeding would have been different.” Strickland, 466 U. S., at 694. In assessing prejudice, courts “must consider the totality of the evidence before the judge or jury.” Id., at 695. The Court of Appeals, however, neglected to take into account the other evidence presented against Thompkins.

The Court of Appeals determined that the state court was unreasonable, 28 U. S. C. § 2254(d), when it found that Thomp-kins suffered no prejudice from failure of defense counsel to request an instruction regarding Purifoy’s earlier acquittal of the murder and assault charges. The state court had rejected Thompkins’ claim that he was prejudiced by evidence of Purifoy’s earlier conviction for firearms offenses, noting that “the record does not disclose an attempt to argue *390that conviction for an improper purpose.” App. to Pet. for Cert. 80a. It is unclear what prejudice standard the state court applied. The Court of Appeals ruled that the state court used the incorrect standard for assessing prejudice under Strickland because “[questions of the prosecution’s purpose or intent are completely irrelevant in ... analyzing whether an error resulted in prejudice, which by definition concerns the error’s effect upon the outcome.” 547 F. 3d, at 591-592 (emphasis deleted).

Even if the state court used an incorrect legal standard, we need not determine whether AEDPA's deferential standard of review, 28 U. S. C. § 2254(d), applies in this situation. Cf. Williams v. Taylor, 529 U. S. 362, 397-398 (2000). That is because, even if AEDPA deference does not apply, Thomp-kins cannot show prejudice under de novo review, the more favorable standard of review for Thompkins. Courts cannot grant writs of habeas corpus under § 2254 by engaging only in de novo review when it is unclear whether AEDPA deference applies, § 2254(d). In those situations, courts must resolve whether AEDPA deference applies, because if it does, a habeas petitioner may not be entitled to a writ of habeas corpus under § 2254(d). Courts can, however, deny writs of habeas corpus under §2254 by engaging in de novo review when it is unclear whether AEDPA deference applies, because a habeas petitioner will not be entitled to a writ of habeas corpus if his or her claim is rejected on de novo review, see § 2254(a).

It seems doubtful that failure to request the instruction about the earlier acquittal or conviction was deficient representation; but on the assumption that it was, on this record Thompkins cannot show prejudice. The record establishes that it was not reasonably likely that the instruction would have made any difference in light of all the other evidence of guilt. The surviving victim, Frederick France, identified Thompkins as the shooter, and the identification was supported by a photograph taken from a surveillance camera. *391Thompkins’ friend Omar Stephens testified that Thompkins confessed to him during a phone conversation, and the details of that confession were corroborated by evidence that Thompkins stripped the van and abandoned it after the shooting. The jury, moreover, was capable of assessing Puri-foy’s credibility, as it was instructed to do. The jury in Thompkins’ case could have concluded that the earlier jury in Purifoy’s case made a mistake, or alternatively, that Puri-foy was not in fact guilty of the crime for which he had been charged. There was ample evidence in the record to support Thompkins’ guilt under either theory, and his jury was instructed to weigh all of the evidence in determining whether there was guilt beyond a reasonable doubt. Under our de novo review of this record, Thompkins cannot show prejudice.

* * *

The judgment of the Court of Appeals is reversed, and the case is remanded with instructions to deny the petition.

It is so ordered.

Justice Sotomayor,

with whom Justice Stevens, Justice Ginsburg, and Justice Breyer join, dissenting.

The Court concludes today that a criminal suspect waives his right to remain silent if, after sitting tacit and uncommunicative through nearly three hours of police interrogation, he utters a few one-word responses. The Court also concludes that a suspect who wishes to guard his right to remain silent against such a finding of “waiver” must, counterintuitively, speak — and must do so with sufficient precision to satisfy a clear-statement rule that construes ambiguity in favor of the police. Both propositions mark a substantial retreat from the protection against compelled self-incrimination that Miranda v. Arizona, 384 U. S. 436 (1966), has long provided during custodial interrogation. The broad rules the Court announces today are also trou*392bling because they are unnecessary to decide this case, which is governed by the deferential standard of review set forth in the Antiterrorism and Effective Death Penalty Act of 1996 (AEDPA), 28 U. S. C. § 2254(d). Because I believe Thomp-kins is entitled to relief under AEDPA on the ground that his statements were admitted at trial without the prosecution having carried its burden to show that he waived his right to remain silent; because longstanding principles of judicial restraint counsel leaving for another day the questions of law the Court reaches out to decide; and because the Court’s answers to those questions do not result from a faithful application of our prior decisions, I respectfully dissent.

I

We granted certiorari to review the judgment of the Court of Appeals for the Sixth Circuit, which held that Thompkins was entitled to habeas relief under both Miranda and Strickland v. Washington, 466 U. S. 668 (1984). 547 F. 3d 572 (2008). As to the Miranda claims, Thompkins argues first that through his conduct during the 3-hour custodial interrogation he effectively invoked his right to remain silent, requiring police to cut off questioning in accordance with Miranda and Michigan v. Mosley, 423 U. S. 96 (1975). Thomp-kins also contends his statements were in any case inadmissible because the prosecution failed to meet its heavy burden under Miranda of proving that he knowingly and intelligently waived his right to remain silent. The Sixth Circuit agreed with Thompkins as to waiver and declined to reach the question of invocation. 547 F. 3d, at 583-584, n. 4. In my view, even if Thompkins cannot prevail on his invocation claim under AEDPA, he is entitled to relief as to waiver. Because I would affirm the judgment of the Sixth Circuit on that ground, I would not reach Thompkins’ claim that he received constitutionally ineffective assistance of counsel.

The strength of Thompkins’ Miranda claims depends in large part on the circumstances of the 3-hour interrogation, *393at the end of which he made inculpatory statements later introduced at trial. The Court’s opinion downplays record evidence that Thompkins remained almost completely silent and unresponsive throughout that session. One of the interrogating officers, Detective Helgert, testified that although Thompkins was administered Miranda warnings, the last of which he read aloud, Thompkins expressly declined to sign a written acknowledgment that he had been advised of and understood his rights. There is conflicting evidence in the record about whether Thompkins ever verbally confirmed understanding his rights.1 The record contains no indication that the officers sought or obtained an express waiver.

As to the interrogation itself, Helgert candidly characterized it as “very, very one-sided” and “nearly a monologue.” App. 10a, 17a. Thompkins was “[pjeculiar,” “[s]ullen,” and “[generally quiet.” Id., at 149a. Helgert and his partner “did most of the talking,” as Thompkins was “not verbally communicative” and “[l]argely” remained silent. Id., at 149a, 17a, 19a. To the extent Thompkins gave any response, his answers consisted of “a word or two. A ‘yeah,’ or a ‘no,’ or T don’t know.’... And sometimes ... he simply sat down ... with [his] head in [his] hands looking down. Sometimes ... he would look up and make eye-contact would be the only response.” Id., at 23a-24a. After proceeding in this fashion for approximately 2 hours and 45 minutes, Helgert *394asked Thompkins three questions relating to his faith in God. The prosecution relied at trial on Thompkins’ one-word answers of “yes.” See id., at 10a-lla.

Thompkins’ nonresponsiveness is particularly striking in the context of the officers’ interview strategy, later explained as conveying to Thompkins that “this was his opportunity to explain his side [of the story]” because “[everybody else, including [his] co-[d]efendants, had given their version,” and asking him “[w]ho is going to speak up for you if you don’t speak up for yourself?” Id., at 10a, 21a. Yet, Helgert confirmed that the “only thing [Thompkins said] relative to his involvement [in the shooting]” occurred near the end of the interview — i.e., in response to the questions about God. Id., at 10a-lla (emphasis added). The only other responses Helgert could remember Thompkins giving were that “ ‘[h]e didn’t want a peppermint’ ” and “ The chair that he was sitting in was hard.’ ” Id., at 152a. Nevertheless, the Michigan court concluded on this record that Thompkins had not invoked his right to remain silent because “he continued to talk with the officer, albeit sporadically,” and that he voluntarily waived that right, People v. Thompkins, No. 242478, (Feb. 3,2004), App. to Pet. for Cert. 75a.

Thompkins’ federal habeas petition is governed by AEDPA, under which a federal court may not grant the writ unless the state court’s adjudication of the merits of the claim at issue “was contrary to, or involved an unreasonable application of, clearly established Federal law, as determined by the Supreme Court of the United States,” or “was based on an unreasonable determination of the facts in light of the evidence presented in the State court proceeding.” §§ 2254(d)(1), (2).

The relevant clearly established federal law for purposes of § 2254(d)(1) begins with our landmark Miranda decision, which “g[a]ve force to the Constitution’s protection against compelled self-incrimination” by establishing “ ‘certain procedural safeguards that require police to advise criminal sus*395pects of their rights under the Fifth and Fourteenth Amendments before commencing custodial interrogation,’ ” Florida v. Powell, 559 U. S. 50, 59 (2010) (quoting Duckworth v. Eagan, 492 U. S. 195, 201 (1989)). Miranda prescribed the now-familiar warnings that police must administer prior to questioning. See 384 U. S., at 479; ante, at 380. Miranda and our subsequent cases also require police to “respect the accused’s decision to exercise the rights outlined in the warnings.” Moran v. Burbine, 475 U. S. 412, 420 (1986). “If [an] individual indicates in any manner, at any time prior to or during questioning, that he -wishes to remain silent” or if he “states that he wants an attorney,” the interrogation “must cease.” 384 U. S., at 473-474.

Even when warnings have been administered and a suspect has not affirmatively invoked his rights, statements made in custodial interrogation may not be admitted as part of the prosecution’s case in chief “unless and until” the prosecution demonstrates that an individual “knowingly and intelligently waive[d] [his] rights.” Id., at 479; accord, ante, at 382. “[A] heavy burden rests on the government to demonstrate that the defendant knowingly and intelligently waived his privilege against self-incrimination and his right to retained or appointed counsel.” Miranda, 384 U. S., at 475. The government must satisfy the “high standard] of proof for the waiver of constitutional rights [set forth in] Johnson v. Zerbst, 304 U. S. 458 (1938).” Ibid.

The question whether a suspect has validly waived his right is “entirely distinct” as a matter of law from whether he invoked that right. Smith v. Illinois, 469 U. S. 91, 98 (1984) (per curiam). The questions are related, however, in terms of the practical effect on the exercise of a suspect’s rights. A suspect may at any time revoke his prior waiver of rights — or, closer to the facts of this case, guard against the possibility of a future finding that he implicitly waived his rights — by invoking the rights and thereby requiring the police to cease questioning. Accord, ante, at 387-388.

*396II

A

Like the Sixth Circuit, I begin with the question whether Thompkins waived his right to remain silent. Even if Thompkins did not invoke that right, he is entitled to relief because Michigan did not satisfy its burden of establishing waiver.

Miranda's discussion of the prosecution’s burden in proving waiver speaks with particular clarity to the facts of this ease and therefore merits reproducing at length:

“If [an] interrogation continues without the presence of an attorney and a statement is taken, a heavy burden rests on the government to demonstrate that the defendant knowingly and intelligently waived his privilege against self-incrimination and his right to retained or appointed counsel.... Since the State is responsible for establishing the isolated circumstances under which [an] interrogation takes place and has the only means of making available corroborated evidence of warnings given during incommunicado interrogation, the burden is rightly on its shoulders.

“An express statement that the individual is willing to make a statement and does not want an attorney followed closely by a statement could constitute a waiver. But a valid waiver will not be presumed simply from the silence of the accused after warnings are given or simply from the fact that a confession was in fact eventually obtained.” 384 U. S., at 475.

Miranda went further in describing the facts likely to satisfy the prosecution’s burden of establishing the admissibility of statements obtained after a lengthy interrogation:

“Whatever the testimony of the authorities as to waiver of rights by an accused, the fact of lengthy interrogation or incommunicado incarceration before a state*397ment is made is strong evidence that the accused did not validly waive his rights. In these circumstances the fact that the individual eventually made a statement is consistent with the conclusion that the compelling influence of the interrogation finally forced him to do so. It is inconsistent with any notion of a voluntary relinquishment of the privilege.” Id., at 476.

This Court’s decisions subsequent to Miranda have emphasized the prosecution’s “heavy burden” in proving waiver. See, e. g., Tague v. Louisiana, 444 U. S. 469, 470-471 (1980) (per curiam); Fare v. Michael C., 442 U. S. 707, 724 (1979). We have also reaffirmed that a court may not presume waiver from a suspect’s silence or from the mere fact that a confession was eventually obtained. See North Carolina v. Butler, 441 U. S. 369, 373 (1979).

Even in concluding that Miranda does not invariably require an express waiver of the right to silence or the right to counsel, this Court in Butler made clear that the prosecution bears a substantial burden in establishing an implied waiver. The Federal Bureau of Investigation had obtained statements after advising Butler of his rights and confirming that he understood them. When presented with a written waiver-of-rights form, Butler told the agents, “ ‘I will talk to you but I am not signing any form.’” 441 U. S., at 371. He then made inculpatory statements, which he later sought to suppress on the ground that he had not expressly waived his right to counsel.

Although this Court reversed the state-court judgment concluding that the statements were inadmissible, we quoted at length portions of the Miranda opinion reproduced above. We cautioned that even an “express written or oral statement of waiver of the right to remain silent or of the right to counsel” is not “inevitably... sufficient to establish waiver,” emphasizing that “[t]he question is... whether the defendant in fact knowingly and voluntarily waived the rights delineated in the Miranda case.” 441 U. S., at 373. Miranda, *398we observed, “unequivocally said . . . mere silence is not enough.” 441 U. S., at 373. While we stopped short in Butler of announcing a per se rule that “the defendant’s silence, coupled with an understanding of his rights and a course of conduct indicating waiver, may never support a conclusion that a defendant has waived his rights,” we reiterated that “courts must presume that a defendant did not waive his rights; the prosecution’s burden is great.” Ibid. 2

Rarely do this Court’s precedents provide clearly established law so closely on point with the facts of a particular case. Together, Miranda and Butler establish that a court “must presume that a defendant did not waive his rights”; the prosecution bears a “heavy burden” in attempting to demonstrate waiver; the fact of a “lengthy interrogation” prior to obtaining statements is “strong evidence” against a finding of valid waiver; “mere silence” in response to questioning is “not enough”; and waiver may not be presumed “simply from the fact that a confession was in fact eventually obtained.” Miranda, supra, at 475-476; Butler, supra, at 372-373.2 3****8

*399It is undisputed here that Thompkins never expressly waived his right to remain silent. His refusal to sign even an acknowledgment that he understood his Miranda rights evinces, if anything, an intent not to waive those rights. Cf. United States v. Plugh, 576 F. 3d 135, 142 (CA2 2009) (suspect’s refusal to sign waiver-of-rights form “constituted an unequivocally negative answer to the question . . . whether he was willing to waive his rights”). That Thompkins did not make the inculpatory statements at issue until after approximately 2 hours and 45 minutes of interrogation serves as “strong evidence” against waiver. Miranda and Butler expressly preclude the possibility that the inculpatory statements themselves are sufficient to establish waiver.

In these circumstances, Thompkins’ “actions and words” preceding the inculpatory statements simply do not evidence a “course of conduct indicating waiver” sufficient to carry the prosecution’s burden. See Butler, supra, at 373.* *4 Al*400though the Michigan court stated that Thompkins “sporadically” participated in the interview, App. to Pet. for Cert. 75a, that court’s opinion and the record before us are silent as to the subject matter or context of even a single question to which Thompkins purportedly responded, other than the exchange about God and the statements respecting the peppermint and the chair. Unlike in Butler, Thompkins made no initial declaration akin to “I will talk to you.” See also 547 F. 3d, at 586-587 (case below) (noting that the case might be different if the record showed Thompkins had responded affirmatively to an invitation to tell his side of the story or described any particular question that Thompkins answered). Indeed, Michigan and the United States concede that no waiver occurred in this case until Thompkins responded “yes” to the questions about God. See Tr. of Oral Arg. 7,30. I believe it is objectively unreasonable under our clearly established precedents to conclude the prosecution met its “heavy burden” of proof on a record consisting of three one-word answers, following 2 hours and 45 minutes of silence punctuated by a few largely nonverbal responses to unidentified questions.

B

Perhaps because our prior Miranda precedents so clearly favor Thompkins, the Court today goes beyond AEDPA’s deferential standard of review and announces a new general principle of law. Any new rule, it must be emphasized, is unnecessary to the disposition of this case. If, in the Court’s view, the Michigan court did not unreasonably apply our Miranda precedents in denying Thompkins relief, it should simply say so and reverse the Sixth Circuit’s judgment on that ground. “It is a fundamental rule of judicial restraint. . . that this Court will not reach constitutional questions in advance of the necessity of deciding them.” Three Affiliated Tribes of Fort Berthold Reservation v. Wold Engineering, P. C., 467 U. S. 138, 157 (1984). Consistent with that rule, we have frequently declined to address questions beyond *401what is necessary to resolve a case under AEDPA. See, e. g., Tyler v. Cain, 533 U. S. 656, 667-668 (2001) (declining to address question where any statement by this Court would be “dictum” in light of AEDPA’s statutory constraints on ha-beas review); cf. Wiggins v. Smith, 539 U. S. 510, 522 (2003) (noting that Williams v. Taylor, 529 U. S. 362 (2000), “made no new law” because the “case was before us on habeas review”). No necessity exists to justify the Court’s broad announcement today.

The Court concludes that when Miranda warnings have been given and understood, “an accused’s uncoerced statement establishes an implied waiver of the right to remain silent.” Ante, at 384. More broadly still, the Court states that, “[a]s a general proposition, the law can presume that an individual who, with a full understanding of his or her rights, acts in a manner inconsistent with their exercise has made a deliberate choice to relinquish the protection those rights afford.” Ante, at 385.

These principles flatly contradict our longstanding views that “a valid waiver will not be presumed . . . simply from the fact that a confession was in fact eventually obtained,” Miranda, 384 U. S., at 475, and that “[t]he courts must presume that a defendant did not waive his rights,” Butler, 441 U. S., at 373. Indeed, we have in the past summarily reversed a state-court decision that inverted Miranda’s antiwaiver presumption, characterizing the error as “readily apparent.” Tague, 444 U. S., at 470-471. At best, the Court today creates an unworkable and conflicting set of presumptions that will undermine Miranda’s goal of providing “concrete constitutional guidelines for law enforcement agencies and courts to follow,” 384 U. S., at 442. At worst, it overrules sub silentio an essential aspect of the protections Miranda has long provided for the constitutional guarantee against self-incrimination.

The Court’s conclusion that Thompkins’ inculpatory statements were sufficient to establish an implied waiver, ante, at *402386-387, finds no support in Butler. Butler itself distinguished between a sufficient “course of conduct” and inculpa-tory statements, reiterating Miranda’s admonition that “ ‘a valid waiver will not be presumed simply from ... the fact that a confession was in fact eventually obtained.’” 441 U. S., at 373 (quoting Miranda, supra, at 475). Michigan suggests Butler’s silence “ Vhen advised of his right to the assistance of a lawyer,’ ” combined with our remand for the state court to apply the implied-waiver standard, shows that silence followed by statements can be a “‘course of conduct.’” Brief for Petitioner 26 (quoting Butler, supra, at 371). But the evidence of implied waiver in Butler was worlds apart from the evidence in this case, because Butler unequivocally said “I will talk to you” after having been read Miranda warnings. Thompkins, of course, made no such statement.

The Court also relies heavily on Burbine in characterizing the scope of the prosecution’s burden in proving waiver. Consistent with Burbine, the Court observes, the prosecution must prove that waiver was “ ‘voluntary in the sense that it was the product of a free and deliberate choice rather than intimidation’” and “ ‘made with a full awareness of both the nature of the right being abandoned and the consequences of the decision to abandon it.’” Ante, at 382-383 (quoting 475 U. S., at 421). I agree with the Court’s statement, so far as it goes. What it omits, however, is that the prosecution also bears an antecedent burden of showing there was, in fact, either an express waiver or a “course of conduct” sufficiently clear to support a finding of implied waiver. Nothing in Burbine even hints at removing that obligation. The question in that case, rather, was whether a suspect’s multiple express waivers of his rights were invalid because police “misinformed an inquiring attorney about their plans concerning the suspect or because they failed to inform the suspect of the attorney’s efforts to reach him.” Id., at 420; see also Colorado v. Spring, 479 U. S. 564, 573 *403(1987). The Court’s analysis in Burbine was predicated on the existence of waiver in fact.

Today’s dilution of the prosecution’s burden of proof to the bare fact that a suspect made inculpatory statements after Miranda warnings were given and understood takes an unprecedented step away from the “high standards of proof for the waiver of constitutional rights” this Court has long demanded. Miranda, supra, at 475; cf. Brewer v. Williams, 430 U. S. 387, 404 (1977) (“[CJourts indulge in every reasonable presumption against waiver”); Zerbst, 304 U. S., at 464. When waiver is to be inferred during a custodial interrogation, there are sound reasons to require evidence beyond inculpatory statements themselves. Miranda and our subsequent cases are premised on the idea that custodial interrogation is inherently coercive. See 384 U. S., at 455 (“Even without employing brutality, the ‘third degree’ or [other] specific strategems . .. the very fact of custodial interrogation exacts a heavy toll on individual liberty and trades on the weakness of individuals”); Dickerson v. United States, 530 U. S. 428, 435 (2000). Requiring proof of a course of conduct beyond the inculpatory statements themselves is critical to ensuring that those statements are voluntary admissions and not the dubious product of an overborne will.

Today’s decision thus ignores the important interests Miranda safeguards. The underlying constitutional guarantee against self-incrimination reflects “many of our fundamental values and most noble aspirations,” our society’s “preference for an accusatorial rather than an inquisitorial system of criminal justice”; a “fear that self-incriminating statements will be elicited by inhumane treatment and abuses” and a resulting “distrust of self-deprecatory statements”; and a realization that while the privilege is “sometimes a shelter to the guilty, [it] is often a protection to the innocent.” Wi-throw v. Williams, 507 U. S. 680, 692 (1993) (internal quotation marks omitted). For these reasons, we have observed, a criminal law system “which comes to depend on the ‘confes*404sion’ will, in the long run, be less reliable and more subject to abuses than a system relying on independent investigation.” Ibid, (some internal quotation marks omitted). “By bracing against ‘the possibility of unreliable statements in every instance of in-custody interrogation/ ” Miranda’s prophylactic rules serve to “ ‘protect the fairness of the trial itself.’ ” 507 U. S., at 692 (quoting Johnson v. New Jersey, 384 U. S. 719, 730 (1966); Schneckloth v. Bustamonte, 412 U. S. 218, 240 (1973)). Today’s decision bodes poorly for the fundamental principles that Miranda protects.

Ill

Thompkins separately argues that his conduct during the interrogation invoked his right to remain silent, requiring police to terminate questioning. Like the Sixth Circuit, I would not reach this question because Thompkins is in any case entitled to relief as to waiver. But even if Thompkins would not prevail on his invocation claim under AEDPA’s deferential standard of review, I cannot agree with the Court’s much broader ruling that a suspect must clearly invoke his right to silence by speaking. Taken together with the Court’s reformulation of the prosecution’s burden of proof as to waiver, today’s novel clear-statement rule for invocation invites police to question a suspect at length — notwithstanding his persistent refusal to answer questions — in the hope of eventually obtaining a single inculpatory response which will suffice to prove waiver of rights. Such a result bears little semblance to the “fully effective” prophylaxis, 384 U. S., at 444, that Miranda requires.

A

Thompkins’ claim for relief under AEDPA rests on the clearly established federal law of Miranda and Mosley. In Miranda, the Court concluded that “[i]f [an] individual indicates in any manner, at any time prior to or during questioning, that he wishes to remain silent, the interrogation must *405cease.... [A]ny statement taken after the person invokes his privilege cannot be other than the product of compulsion, subtle or otherwise.” 384 U. S., at 473-474. In Mosley, the Court said that a “critical safeguard” of the right to remain silent is a suspect’s “‘right to cut off questioning.’” 423 U. S., at 103 (quoting Miranda, supra, at 474). Thus, “the admissibility of statements obtained after the person in custody has decided to remain silent depends under Miranda on whether his ‘right to cut off questioning’ was ‘scrupulously honored.’” 423 U. S., at 104.5

Thompkins contends that in refusing to respond to questions he effectively invoked his right to remain silent, such that police were required to terminate the interrogation prior to his inculpatory statements. In Michigan’s view, Thompkins cannot prevail under AEDPA because this Court’s precedents have not previously established whether a suspect’s ambiguous statements or actions require the police to stop questioning. We have held that a suspect who has “‘invoked his right to have counsel present ... is not subject to farther interrogation by the authorities until counsel has been made available to him, unless [he] initiates further communication, exchanges, or conversations with the police.’” Maryland v. Shatzer, 559 U. S. 98, 104 (2010) (quoting Edwards v. Arizona, 451 U. S. 477, 484-485 (1981)). Notwithstanding Miranda’s statement that “there can be no questioning” if a suspect “indicates in any manner . . . that he wishes to consult with an attorney,” 384 U. S., at 444-445, the Court in Davis v. United States, 512 U. S. 452, 461 (1994), *406established a clear-statement rule for invoking the right to counsel. After a suspect has knowingly and voluntarily waived his Miranda rights, Davis held, police may continue questioning “until and unless the suspect clearly requests an attorney.” 512 U. S., at 461 (emphasis added).

Because this Court has never decided whether Davis' clear-statement rule applies to an invocation of the right to silence, Michigan contends, there was no clearly established federal law prohibiting the state court from requiring an unambiguous invocation. That the state court’s decision was not objectively unreasonable is confirmed, in Michigan’s view, by the number of Federal Courts of Appeals to have applied Davis to invocation of the right to silence. Brief for Petitioner 44.

Under AJEDPA’s deferential standard of review, it is indeed difficult to conclude that the state court’s application of our precedents was objectively unreasonable. Although the duration and consistency of Thompkins’ refusal to answer questions throughout the 3-hour interrogation provide substantial evidence in support of his claim, Thompkins did not remain absolutely silent, and this Court has not previously addressed whether a suspect can invoke the right to silence by remaining uncooperative and nearly silent for 2 hours and 45 minutes.

B

The Court, however, eschews this narrow ground of decision, instead extending Davis to hold that police may continue questioning a suspect until he unambiguously invokes his right to remain silent. Because Thompkins neither said “he wanted to remain silent” nor said “he did not want to talk with the police,” the Court concludes, he did not clearly invoke his right to silence. Ante, at 380-382.6

*407I disagree with this novel application of Davis. Neither the rationale nor holding of that case compels today’s result. Davis involved the right to counsel, not the right to silence. The Court in Davis reasoned that extending Edwards’ “rigid” prophylactic rule to ambiguous requests for a lawyer would transform Miranda into a “ ‘wholly irrational obstacle] to legitimate police investigative activity’” by “needlessly preventing] the police from questioning a suspect in the absence of counsel even if [he] did not wish to have a lawyer present.” Davis, supra, at 460. But Miranda itself “distinguished between the procedural safeguards triggered by a request to remain silent and a request for an attorney.” Mosley, 423 U. S., at 104, n. 10; accord, Edwards, supra, at 485. Mosley upheld the admission of statements when police immediately stopped interrogating a suspect who invoked his right to silence, but reapproached him after a 2-hour delay and obtained inculpatory responses relating to a different crime after administering fresh Miranda warnings. The different effects of invoking the rights are consistent with distinct standards for invocation. To the extent Mosley contemplates a more flexible form of prophylaxis than Edwards — and, in particular, does not categorically bar police from reapproaching a suspect who has invoked his right to remain silent — Davis’ concern about “‘wholly irrational obstacles’ ” to police investigation applies with less force.

In addition, the suspect’s equivocal reference to a lawyer in Davis occurred only after he had given express oral and written waivers of his rights. Davis’ holding is explicitly predicated on that fact. See 512 U. S., at 461 (“We therefore hold that, after a knowing and voluntary waiver of the Miranda rights, law enforcement officers may continue questioning until and unless the suspect clearly requests an attorney”). The Court ignores this aspect of Davis, as well as the decisions of numerous federal and state courts declining *408to apply a dear-statement rule when a suspect has not previously given an express waiver of rights.7

In my mind, a more appropriate standard for addressing a suspect’s ambiguous invocation of the right to remain silent is the constraint Mosley places on questioning a suspect who has invoked that right: The suspect’s “ ‘right to cut off questioning’ ” must be “ ‘scrupulously honored.’ ” See 423 U. S., at 104. Such a standard is necessarily precautionary and fact specific. The rule would acknowledge that some statements or conduct are so equivocal that police may scrupulously honor a suspect’s rights without terminating questioning — for instance, if a suspect’s actions are reasonably understood to indicate a willingness to listen before deciding whether to respond. But other statements or actions — in particular, when a suspect sits silent throughout prolonged interrogation, long past the point when he could be deciding whether to respond — cannot reasonably be understood other than as an invocation of the right to remain silent. Under such circumstances, “scrupulous” respect for the suspect's rights will require police to terminate questioning under Mosley. 8

*409To be sure, such a standard does not provide police with a bright-line rule. Cf. ante, at 381-382. But, as we have previously recognized, Mosley itself does not offer clear guidance to police about when and how interrogation may continue after a suspect invokes his rights. See Solem v. Stumes, 465 U. S. 638, 648 (1984); see also Shatzer, 559 U. S., at 119 (Thomas, J., concurring in part and concurring in judgment). Given that police have for nearly 35 years applied Mosley’s fact-specific standard in questioning suspects who have invoked their right to remain silent; that our cases did not during that time resolve what statements or actions suffice to invoke that right; and that neither Michigan nor the Solicitor General has provided evidence in this case that the status quo has proved unworkable, I see little reason to believe today’s clear-statement rule is necessary to ensure effective law enforcement.

Davis’ clear-statement rule is also a poor fit for the right to silence. Advising a suspect that he has a “right to remain silent” is unlikely to convey that he must speak (and must do so in some particular fashion) to ensure the right will be protected. Cf. Soffar v. Cockrell, 300 F. 3d 588, 603 (CA5 2002) (en banc) (DeMoss, J., dissenting) (“What in the world must an individual do to exercise his constitutional right to remain silent beyond actually, in fact, remaining silent?”). By contrast, telling a suspect “he has the right to the presence of an attorney, and that if he cannot afford an attorney one will be appointed for him prior to any questioning if he so desires,” Miranda, 384 U. S., at 479, implies the need for speech to exercise that right. Davis’ requirement that a suspect must “clearly requesft] an attorney” to terminate questioning thus aligns with a suspect’s likely understanding of the Miranda warnings in a way today’s rule does not. 512 U. S., at 461. The Court suggests Thompkins could have employed the “simple, unambiguous” means of saying “he wanted to remain silent” or “did not want to talk with the police.” Ante, at 382. But the Miranda warnings give no *410hint that a suspect should use those magic words, and there is little reason to believe police — who have ample incentives to avoid invocation — will provide such guidance.

Conversely, the Court’s concern that police will face “difficult decisions about an accused’s unclear intent” and suffer the consequences of “ ‘guess[ing] wrong,’ ” ante, at 382 (quoting Davis, 512 U. S., at 461), is misplaced. If a suspect makes an ambiguous statement or engages in conduct that creates uncertainty about his intent to invoke his right, police can simply ask for clarification. See id., at 467 (Souter, J., concurring in judgment). It is hardly an unreasonable burden for police to ask a suspect, for instance, “Do you want to talk to us?” The majority in Davis itself approved of this approach as protecting suspects’ rights while “minimiz[ing] the chance of a confession [later] being suppressed.” Id., at 461. Given this straightforward mechanism by which police can “scrupulously hono[r]” a suspect's right to silence, today’s clear-statement rule can only be seen as accepting “as tolerable the certainty that some poorly expressed requests [to remain silent] will be disregarded,” id., at 471 (opinion of Souter, J.), without any countervailing benefit. Police may well prefer not to seek clarification of an ambiguous statement out of fear that a suspect will invoke his rights. But “our system of justice is not founded on a fear that a suspect will exercise his rights. ‘If the exercise of constitutional rights will thwart the effectiveness of a system of law enforcement, then there is something very wrong with that system.’” Burbine, 475 U. S., at 458 (Stevens, J., dissenting) (quoting Escobedo v. Illinois, 378 U. S. 478, 490 (1964)).

The Court asserts in passing that treating ambiguous statements or acts as an invocation of the right to silence will only “‘marginally’” serve Miranda’s goals. Ante, at 382. Experience suggests the contrary. In the 16 years sinee Davis was decided, ample evidence has accrued that criminal suspects often use equivocal or colloquial language in attempting to invoke their right to silence. A number of *411lower courts that have (erroneously, in my view) imposed a clear-statement requirement for invocation of the right to silence have rejected as ambiguous an array of statements whose meaning might otherwise be thought plain.9 At a minimum, these decisions suggest that differentiating “clear” from “ambiguous” statements is often a subjective inquiry. Even if some of the cited decisions are themselves in tension with Davis’ admonition that a suspect need not “ ‘speak with the discrimination of an Oxford don’” to invoke his rights, *412512 U. S., at 459 (quoting id., at 476 (opinion of Souter, J.)), they demonstrate that today’s decision will significantly burden the exercise of the right to silence. Notably, when a suspect “understands his (expressed) wishes to have been ignored ... in contravention of the ‘rights’ just read to him by his interrogator, he may well see further objection as futile and confession (true or not) as the only way to end his interrogation.” Id., at 472-473.

For these reasons, I believe a precautionary requirement that police “scrupulously hono[r]” a suspect’s right to cut off questioning is a more faithful application of our precedents than the Court’s awkward and needless extension of Davis.

* * *

Today’s decision turns Miranda upside down. Criminal suspects must now unambiguously invoke their right to remain silent — which, counterintuitively, requires them to speak. At the same time, suspects will be legally presumed to have waived their rights even if they have given no clear expression of their intent to do so. Those results, in my view, find no basis in Miranda or our subsequent cases and are inconsistent with the fair-trial principles on which those precedents are grounded. Today's broad new rules are all the more unfortunate because they are unnecessary to the disposition of the case before us. I respectfully dissent.

Justice Powell

delivered the opinion of the Court.

In Miranda v. Arizona, 384 U. S. 436 (1966), the Court held that a suspect's waiver of the Fifth Amendment privilege against self-incrimination is valid only if it is made voluntarily, knowingly, and intelligently. Id., at 444. This case presents the question whether the suspect’s awareness of all the crimes about which he may be questioned is relevant to determining the validity of his decision to waive the Fifth Amendment privilege.

I

In February 1979, respondent John Leroy Spring and a companion shot and killed Donald Walker during a hunting trip in Colorado. Shortly thereafter, an informant told agents of the Bureau of Alcohol, Tobacco, and Firearms (ATF) that Spring was engaged in the interstate transportation of stolen firearms. The informant also told the agents that Spring had discussed his participation in the Colorado killing. At the time the ATF agents received this information, Walker’s body had not been found and the police had received no report of his disappearance. Based on the information received from the informant relating to the firearms violations, the ATF agents set up an undercover operation to purchase firearms from Spring. On March 30, 1979, ATF agents arrested Spring in Kansas City, Missouri, during the undercover purchase.

*567An ATF agent on the scene of the arrest advised Spring of his Miranda rights.1 Spring was advised of his Miranda rights a second time after he was transported to the ATF office in Kansas City. At the ATF office, the agents also advised Spring that he had the right to stop the questioning at any time or to stop the questioning until the presence of an attorney could be secured. Spring then signed a written form stating that he understood and waived his rights, and that he was willing to make a statement and answer questions.

ATF agents first questioned Spring about the firearms transactions that led to his arrest. They then asked Spring if he had a criminal record. He admitted that he had a juvenile record for shooting his aunt when he was 10 years old. The agents asked if Spring had ever shot anyone else. Spring ducked his head and mumbled, “I shot another guy once.” The agents asked Spring if he had ever been to Colorado. Spring said no. The agents asked Spring whether he had shot a man named Walker in Colorado and thrown his body into a snowbank. Spring paused and then ducked his head again and said no. The interview ended at this point.

On May 26, 1979, Colorado law enforcement officials visited Spring while he was in jail in Kansas City pursuant to his arrest on the firearms offenses. The officers gave Spring the Miranda warnings, and Spring again signed a written form indicating that he understood his rights and was willing to waive them. The officers informed Spring that they wanted to question him about the Colorado homicide. Spring indicated that he “wanted to get it off his chest.” In an interview that lasted approximately IV2 hours, Spring confessed to the Colorado murder. During that time, Spring *568talked freely to the officers, did not indicate a desire to terminate the questioning, and never requested counsel. The officers prepared a written statement summarizing the interview. Spring read, edited, and signed the statement.

Spring was charged in Colorado state court with first-degree murder. Spring moved to suppress both statements on the ground that his waiver of Miranda rights was invalid. The trial court found that the ATF agents’ failure to inform Spring before the March 30 interview that they would question him about the Colorado murder did not affect his waiver of his Miranda rights:

“[T]he questions themselves suggested the topic of inquiry. The questions dealt with ‘shooting anyone’ and specifically killing a man named Walker and throwing his body in a snowbank in Colorado. The questions were not designed to gather information relating to a subject that was not readily evident or apparent to Spring. Spring had been advised of his right to remain silent, his right to stop answering questions, and to have an Attorney present during interrogation. He did not elect to exercise his right to remain silent or to refuse to answer questions relating to the homicide, nor did he request Counsel during interrogation.” App. to Pet. for Cert. 4-A.

Accordingly, the trial court concluded that the March 30 statement should not be suppressed on Fifth Amendment grounds. The trial court, however, subsequently ruled that Spring’s statement that he “shot another guy once” was irrelevant, and that the context of the discussion did not support the inference that the statement related to the Walker homicide. For that reason, the March 30 statement was not admitted at Spring’s trial. The court concluded that the May 26 statement “was made freely, voluntarily, and intelligently, after [Spring’s] being properly and fully advised of his rights, and that the statement should not be suppressed, but should *569be admitted in evidence.” Id., at 5-A. The May 26 statement was admitted into evidence at trial, and Spring was convicted of first-degree murder.2

Spring argued on appeal that his waiver of Miranda rights before the March 30 statement was invalid because he was not informed that he would be questioned about the Colorado murder. Although this statement was not introduced at trial, he claimed that its validity was relevant because the May 26 statement that was admitted against him was the illegal “fruit” of the March 30 statement, see Wong Sun v. United States, 371 U. S. 471 (1963), and therefore should have been suppressed. The Colorado Court of Appeals agreed with Spring, holding that the ATF agents “had a duty to inform Spring that he was a suspect, or to readvise him of his Miranda rights, before questioning him about the murder.” 671 P. 2d 965, 966 (1983). Because they failed to do so before the March 30 interview, “any waiver of rights in regard to questions designed to elicit information about Walker’s death was not given knowingly or intelligently.” Id., at 967. The court held that the March 30 statement was inadmissible and that the State had failed to meet its burden of proving that the May 26 statement was not the product of the prior illegal statement. The court reversed Spring’s conviction and remanded the case for a new trial, directing that if the State sought to introduce the May 26 statement into evidence, the trial court should determine whether the “taint” of *570the March 30 statement was sufficiently attenuated to allow introduction of the May 26 statement.

The Colorado Supreme Court affirmed the judgment of the Court of Appeals, although its reasoning differed in some respects. 713 P. 2d 865 (1985). The court found:

“[T]he validity of Spring’s waiver of constitutional rights must be determined upon an examination of the totality of the circumstances surrounding the making of the statement to determine if the waiver was voluntary, knowing and intelligent. No one factor is always determinative in that analysis. Whether, and to what extent, a suspect has been informed or is aware of the subject matter of the interrogation prior to its commencement is simply one factor in the court’s evaluation of the total circumstances, although it may be a major or even a determinative factor in some situations.” Id., at 872-873 (citations omitted).

The court concluded:

“Here, the absence of an advisement to Spring that he would be questioned about the Colorado homicide, and the lack of any basis to conclude that at the time of the execution of the waiver, he reasonably could have expected that the interrogation would extend to that subject, are determinative factors in undermining the validity of the waiver.” Id., at 874 (emphasis in original).

Justice Erickson, joined by Justice Rovira, dissented as to the resolution of this issue, stating:

“Law enforcement officers have no duty under Miranda to inform a person in custody of all charges being investigated prior to questioning him. All that Miranda requires is that the suspect be advised that he has the right to remain silent, that anything he says can and will be used against him in court, that he has the right to consult with a lawyer and to have the lawyer present during interrogation, and that if he cannot afford a law*571yer one will be appointed to represent him.” Id., at 880 (citations omitted).

The dissenting justices found “ample evidence to support the trial court’s conclusion that Spring waived his Miranda rights” and rejected “the majority’s conclusion that Spring’s waiver of his Miranda rights on March 30, 1979 was invalid simply because he was not informed of all matters that would be reviewed when he was questioned by the police.” Id., at 881. The court remanded the case for further proceedings consistent with its opinion.

We granted certiorari, 476 U. S. 1104 (1986), to resolve an arguable Circuit conflict3 and to review the Colorado Supreme Court’s determination that a suspect’s awareness of the possible subjects of questioning is a relevant and sometimes determinative consideration in assessing whether a waiver of the Fifth Amendment privilege is valid. We now reverse.

II

There is no dispute that the police obtained the May 26 confession after complete Miranda warnings and after informing Spring that he would be questioned about the Colorado homicide. The Colorado Supreme Court nevertheless held that the confession should have been suppressed because it was the illegal “fruit” of the March 30 statement. A confession cannot be “fruit of the poisonous tree” if the tree itself is not *572poisonous. Our inquiry, therefore, centers on the validity of the March 30 statement.4

A

The Fifth Amendment of the United States Constitution provides that no person “shall be compelled in any criminal case to be a witness against himself.”5 This privilege “is fully applicable during a period of custodial interrogation.” Miranda v. Arizona, 384 U. S., at 460-461.6 In Miranda, the Court concluded that “without proper safeguards the process of in-custody interrogation of persons suspected or accused of crime contains inherently compelling pressures which work to undermine the individual’s will to resist and to compel him to speak where he would not otherwise do so freely.” Id., at 467. Accordingly, the Court formulated the now-familiar “procedural safeguards effective to secure the privilege against self-incrimination.” Id., at 444. The Court’s fundamental aim in designing the Miranda warnings was “to assure that the individual’s right to choose between silence and speech remains unfettered throughout the interrogation process.” Id., at 469.

Consistent with this purpose, a suspect may waive his Fifth Amendment privilege, “provided the waiver is made voluntarily, knowingly and intelligently.” Id., at 444. In this case, the law enforcement officials twice informed Spring *573of his Fifth Amendment privilege in precisely the manner specified by Miranda. As we have noted, Spring indicated that he understood the enumerated rights and signed a written form expressing his intention to waive his Fifth Amendment privilege. The trial court specifically found that “there was no element of duress or coercion used to induce Spring’s statements [on March 30, 1978].” App. to Pet. for Cert. 3-A. Despite the explicit warnings and the finding by the trial court, Spring argues that his March 30 statement was in effect compelled in violation of his Fifth Amendment privilege because he signed the waiver form without being aware that he would be questioned about the Colorado homicide. Spring’s argument strains the meaning of compulsion past the breaking point.

B

A statement is not “compelled” within the meaning of the Fifth Amendment if an individual “voluntarily, knowingly and intelligently” waives his constitutional privilege. Miranda v. Arizona, supra, at 444. The inquiry whether a waiver is coerced “has two distinct dimensions.” Moran v. Burbine, 475 U. S. 412, 421 (1986):

“First the relinquishment of the right must have been voluntary in the sense that it was the product of a free and deliberate choice rather than intimidation, coercion, or deception. Second, the waiver must have been made with a full awareness both of the nature of the right being abandoned and the consequences of the decision to abandon it. Only if the ‘totality of the circumstances surrounding the interrogation’ reveal both an uncoerced choice and the requisite level of comprehension may a court properly conclude that the Miranda rights have been waived.” Ibid, (quoting Fare v. Michael C., 442 U. S. 707, 725 (1979)).

There is no doubt that Spring’s decision to waive his Fifth Amendment privilege was voluntary. He alleges no “coer*574cion of a confession by physical violence or other deliberate means calculated to break [his] will,” Oregon v. Elstad, 470 U. S. 298, 312 (1985), and the trial court found none. His allegation that the police failed to supply him with certain information does not relate to any of the traditional indicia of coercion: “the duration and conditions of detention . . . , the manifest attitude of the police toward him, his physical and mental state, the diverse pressures which sap or sustain his powers of resistance and self-control.” Culombe v. Connecticut, 367 U. S. 568, 602 (1961) (opinion of Frankfurter, J.). Absent evidence that Spring’s “will [was] overborne and his capacity for self-determination critically impaired” because of coercive police conduct, ibid.; see Colorado v. Connelly, 479 U. S. 157, 163-164 (1986), his waiver of his Fifth Amendment privilege was voluntary under this Court’s decision in Miranda.

There also is no doubt that Spring’s waiver of his Fifth Amendment privilege was knowingly and intelligently made: that is, that Spring understood that he had the right to remain silent and that anything he said could be used as evidence against him. The Constitution does not require that a criminal suspect know and understand every possible consequence of a waiver of the Fifth Amendment privilege. Moran v. Burbine, supra, at 422; Oregon v. Elstad, supra, at 316-317. The Fifth Amendment’s guarantee is both simpler and more fundamental: A defendant may not be compelled to be a witness against himself in any respect. The Miranda warnings protect this privilege by ensuring that a suspect knows that he may choose not to talk to law enforcement officers, to talk only with counsel present, or to discontinue talking at any time. The Miranda warnings ensure that a waiver of these rights is knowing and intelligent by requiring that the suspect be fully advised of this constitutional privilege, including the critical advice that whatever he chooses to say may be used as evidence against him.

*575In this case there is no allegation that Spring failed to understand the basic privilege guaranteed by the Fifth Amendment. Nor is there any allegation that he misunderstood the consequences of speaking freely to the law enforcement officials. In sum, we think that the trial court was indisputably correct in finding that Spring’s waiver was made knowingly and intelligently within the meaning of Miranda.

hH b-1

A

Spring relies on this Court’s statement in Miranda that “any evidence that the accused was threatened, tricked, or cajoled into a waiver will. . . show that the defendant did not voluntarily waive his privilege. ” 384 U. S., at 476. He contends that the failure to inform him of the potential subjects of interrogation constitutes the police trickery and deception condemned in Miranda, thus rendering his waiver of Miranda rights invalid. Spring, however, reads this statement in Miranda out of context and without due regard to the constitutional privilege the Miranda warnings were designed to protect.

We note first that the Colorado courts made no finding of official trickery.7 In fact, as noted above, the trial court expressly found that “there was no element of duress or coercion used to induce Spring’s statements.” Supra, at 573. *576Spring nevertheless insists that the failure of the ATF agents to inform him that he would be questioned about the murder constituted official “trickery” sufficient to invalidate his waiver of his Fifth Amendment privilege, even if the official conduct did not amount to “coercion.” Even assuming that Spring’s proposed distinction has merit, we reject his conclusion. This Court has never held that mere silence by law enforcement officials as to the subject matter of an interrogation is “trickery” sufficient to invalidate a suspect’s waiver of Miranda rights, and we expressly decline so to hold today.8

Once Miranda warnings are given, it is difficult to see how official silence could cause a suspect to misunderstand the nature of his constitutional right — “his right to refuse to answer any question which might incriminate him.” United States v. Washington, 431 U. S. 181, 188 (1977). “Indeed, it seems self-evident that one who is told he is free to refuse to answer questions is in a curious posture to later complain that his answers were compelled.” Ibid. We have held that a valid waiver does not require that an individual be informed of all information “useful” in making his decision or all information that “might . . . affec[t] his decision to confess.” Moran v. Burbine, 475 U. S., at 422. “[W]e have never read the Constitution to require that the police supply a suspect with a flow of information to help him calibrate his self-interest in *577deciding whether to speak or stand by his rights.” Ibid. 9 Here, the additional information could affect only the wisdom of a Miranda waiver, not its essentially voluntary and knowing nature. Accordingly, the failure of the law enforcement officials to inform Spring of the subject matter of the interrogation could not affect Spring’s decision to waive his Fifth Amendment privilege in a constitutionally significant manner.

B

This Court’s holding in Miranda specifically required that the police inform a criminal suspect that he has the right to remain silent and that anything he says may be used against him. There is no qualification of this broad and explicit warning. The warning, as formulated in Miranda, conveys to a suspect the nature of his constitutional privilege and the consequences of.abandoning it. Accordingly, we hold that a suspect’s awareness of all the possible subjects of questioning in advance of interrogation is not relevant to determining whether the suspect voluntarily, knowingly, and intelligently waived his Fifth Amendment privilege.

f — I <1

The judgment of the Colorado Supreme Court is reversed, and the case is remanded for further proceedings not inconsistent with this opinion.

It is so ordered.

Justice Marshall,

with whom Justice Brennan joins, dissenting.

The Court asserts there is “no doubt” that respondent Spring’s decision to waive his Fifth Amendment privilege *578was voluntarily, knowingly, and intelligently made. Ante, at 573 and 574. I agree, however, with the Colorado Supreme Court that a significant doubt exists in the circumstances of this case and thus the State has failed to carry the “heavy burden” recognized in Miranda v. Arizona, 384 U. S. 436, 475 (1966), for establishing the constitutional validity of Spring’s alleged waiver.

Consistent with our prior decisions, the Court acknowledges that a suspect’s waiver of fundamental constitutional rights, such as Miranda’s protections against self-incrimination during a custodial interrogation, must be examined in light of the “‘“totality of the circumstances.””’ Ante, at 573, quoting Moran v. Burbine, 475 U. S. 412, 421 (1986), in turn quoting Fare v. Michael C., 442 U. S. 707, 725 (1979); see also id., at 724-725; North Carolina v. Butler, 441 U. S. 369, 374-375 (1979); Johnson v. Zerbst, 304 U. S. 458, 464 (1938). Nonetheless, the Court proceeds to hold that the specific crimes and topics of investigation known to the interrogating officers before questioning begins are “not relevant” to, and in this case “could not affect,” the validity of the suspect’s decision to waive his Fifth Amendment privilege. Ante, at 577. It seems to me self-evident that a suspect’s decision to waive this privilege will necessarily be influenced by his awareness of the scope and seriousness of the matters under investigation.

To attempt to minimize the relevance of such information by saying that it “could affect only the wisdom of” the suspect’s waiver, as opposed to the validity of that waiver, ventures an inapposite distinction. Ibid. Wisdom and validity in this context are overlapping concepts, as circumstances relevant to assessing the validity of a waiver may also be highly relevant to its wisdom in any given context. Indeed, the admittedly “critical” piece of advice the Court recognizes today — that the suspect be informed that whatever he says may be used as evidence against him — is certainly relevant to the wisdom of any suspect’s decision to submit to custodial interrogation without first consulting his lawyer. Ante, at *579574. The Court offers no principled basis for concluding that this is a relevant factor for determining the validity of a waiver but that, under what it calls a totality of the circumstances analysis, a suspect’s knowledge of the specific crimes and other topics previously identified for questioning can never be. -

The Court quotes Moran v. Burbine, supra, at 422, as holding that “a valid waiver does not require that an individual be informed of all information ‘useful’ in making his decision or all information that ‘might . . . affec[t] his decision to confess.’” Ante, at 576 (emphasis added). Noticeably similar is the Court’s holding today: “[A] suspect’s awareness of all the possible subjects of questioning in advance of interrogation is not relevant to determining” the validity of his waiver. Ante, at 577 (emphasis added). This careful phraseology avoids the important question whether the lack of any indication of the identified subjects for questioning is relevant to determining the validity of the suspect’s waiver.

I would include among the relevant factors for consideration whether before waiving his Fifth Amendment rights the suspect was aware, either through the circumstances surrounding his arrest or through a specific advisement from the arresting or interrogating officers, of the crime or crimes he was suspected of committing and about which they intended to ask questions. To hold that such knowledge is relevant would not undermine the “‘virtue of informing police and prosecutors with specificity’ as to how a pretrial questioning of a suspect must be conducted,” ante, at 577, n. 9 (quoting Fare v. Michael C., supra, at 718), nor would it interfere with the use of legitimate interrogation techniques. Indeed, requiring the officers to articulate at a minimum the crime or crimes for which the suspect has been arrested could contribute significantly toward ensuring that the arrest was in fact lawful and the suspect’s statement not compelled because of an error at this stage alone, a problem we addressed in Brown v. Illinois, 422 U. S. 590, 601 (1975), under the *580Fourth Amendment on the assumption that the defendant’s waiver of his Fifth Amendment rights in that case had been voluntary. See also Dunaway v. New York, 442 U. S. 200, 217 (1979) (voluntary waiver of Miranda warnings is a threshold requirement for Fourth Amendment analysis).

The interrogation tactics utilized in this case demonstrate the relevance of the information Spring did not receive. The agents evidently hoped to obtain from Spring a valid confession to the federal firearms charge for which he was arrested and then parlay this admission into an additional confession of first-degree murder. Spring could not have expected questions about the latter, separate offense when he agreed to waive his rights, as it occurred in a different State and was a violation of state law outside the normal investigative focus of federal Alcohol, Tobacco, and Firearms agents.

“Interrogators describe the point of the first admission as the ‘breakthrough’ and the ‘beachhead,’ R. Royal & S. Schutt, The Gentle Art of Interviewing and Interrogation: A Professional Manual and Guide 143 (1976), which once obtained will give them enormous ‘tactical advantages,’ F. Inbau & J. Reid, Criminal Interrogation and Confessions 82 (2d ed. 1967).” Oregon v. Elstad, 470 U. S. 298, 328 (1985) (Brennan, J., dissenting). The coercive aspects of the psychological ploy intended in this case, when combined with an element of surprise which may far too easily rise to a level of deception,1 cannot be justified in light of Miranda’s strict *581requirements that the suspect’s waiver and confession be voluntary, knowing, and intelligent. 384 U. S., at 445-458, 475-476. If a suspect has signed a waiver form with the intention of making a statement regarding a specifically alleged crime, the Court today would hold this waiver valid with respect to questioning about any other crime, regardless of its relation to the charges the suspect believes he will be asked to address. Yet once this waiver is given and the intended statement made, the protections afforded by Miranda against the “inherently compelling pressures” of the custodial interrogation, id., at 467, have effectively dissipated. Additional questioning about entirely separate and more serious suspicions of criminal activity can take unfair advantage of the suspect’s psychological state, as the unexpected questions cause the compulsive pressures suddenly to reappear. Given this technique of interrogation, a suspect’s understanding of the topics planned for questioning is, therefore, at the very least “relevant” to assessing whether his decision to talk to the officers was voluntarily, knowingly, and intelligently made.

Not only is the suspect’s awareness of the suspected criminal conduct relevant, its absence may be determinative in a given case. The State’s burden of proving that a suspect’s waiver was voluntary, knowing, and intelligent is a “heavy” one. Miranda, 384 U. S., at 475. We are to “‘indulge every reasonable presumption against waiver’ of fundamental constitutional rights” and we shall “‘not presume acquiescence in the loss of fundamental rights.’” Johnson, *582304 U. S., at 464 (citations omitted); see Brewer v. Williams, 430 U. S. 387, 404 (1977). It is reasonable to conclude that, had Spring known of the federal agents’ intent to ask questions about a murder unrelated to the offense for which he was arrested, he would not have consented to interrogation without first consulting his attorney. In this case, I would therefore accept the determination of the Colorado Supreme Court that Spring did not voluntarily, knowingly, and intelligently waive his Fifth Amendment rights. 713 P. 2d 865, 873-874 (1985).2

I dissent.

*651Justice Rehnquist

delivered the opinion of the Court.

Respondent Benjamin Quarles was charged in the New York trial court with criminal possession of a weapon. The trial court suppressed the gun in question, and a statement made by respondent, because the statement was obtained by police before they read respondent his “Miranda rights.” That ruling was affirmed on appeal through the New York Court of Appeals. We granted certiorari, 461 U. S. 942 (1983), and we now reverse.1 We conclude that under the circumstances involved in this case, overriding considerations of public safety justify the officer’s failure to provide Miranda warnings before he asked questions devoted to locating the abandoned weapon.

On September 11, 1980, at approximately 12:30 a. m., Officer Frank Kraft and Officer Sal Scarring were on road patrol in Queens, N. Y., when a young woman approached their car. She told them that she had just been raped by a black male, approximately six feet tall, who was wearing a black jacket with the name “Big Ben” printed in yellow letters on the back. She told the officers that the man had just entered *652an A & P supermarket located nearby and that the man was carrying a gun.

The officers drove the woman to the supermarket, and Officer Kraft entered the store while Officer Scarring radioed for assistance. Officer Kraft quickly spotted respondent, who matched the description given by the woman, approaching a checkout counter. Apparently upon seeing the officer, respondent turned and ran toward the rear of the store, and Officer Kraft pursued him with a drawn gun. When respondent turned the corner at the end of an aisle, Officer Kraft lost sight of him for several seconds, and upon regaining sight of respondent, ordered him to stop and put his hands over his head.

Although more than three other officers had arrived on the scene by that time, Officer Kraft was the first to reach respondent. He frisked him and discovered that he was wearing a shoulder holster which was then empty. After handcuffing him, Officer Kraft asked him where the gun was. Respondent nodded in the direction of some empty cartons and responded, “the gun is over there.” Officer Kraft thereafter retrieved a loaded .38-caliber revolver from one of the cartons, formally placed respondent under arrest, and read him his Miranda rights from a printed card. Respondent indicated that he would be willing to answer questions without an attorney present. Officer Kraft then asked respondent if he owned the gun and where he had purchased it. Respondent answered that he did own it and that he had purchased it in Miami, Fla.

In the subsequent prosecution of respondent for criminal possession of a weapon,2 the judge excluded the statement, “the gun is over there,” and the gun because the officer had not given respondent the warnings required by our decision in Miranda v. Arizona, 384 U. S. 436 (1966), before asking *653him where the gun was located. The judge excluded the other statements about respondent’s ownership of the gun and the place of purchase, as evidence tainted by the prior Miranda violation. The Appellate Division of the Supreme Court of New York affirmed without opinion. 85 App. Div. 2d 936, 447 N. Y. S. 2d 84 (1981).

The Court of Appeals granted leave to appeal and affirmed by a 4-3 vote. 58 N. Y. 2d 664, 444 N. E. 2d 984 (1982). It concluded that respondent was in “custody” within the meaning of Miranda during all questioning and rejected the State’s argument that the exigencies of the situation justified Officer Kraft’s failure to read respondent his Miranda rights until after he had located the gun. The court declined to recognize an exigency exception to the usual requirements of Miranda because it found no indication from Officer Kraft’s testimony at the suppression hearing that his subjective motivation in asking the question was to protect his own safety or the safety of the public. 58 N. Y. 2d, at 666, 444 N. E. 2d, at 985. For the reasons which follow, we believe that this case presents a situation where concern for public safety must be paramount to adherence to the literal language of the prophylactic rules enunciated in Miranda.3

*654The Fifth Amendment guarantees that “[n]o person . . . shall be compelled in any criminal case to be a witness against himself.” In Miranda this Court for the first time extended the Fifth Amendment privilege against compulsory self-incrimination to individuals subjected to custodial interrogation by the police. 384 U. S., at 460-461, 467. The Fifth Amendment itself does not prohibit all incriminating admissions; “[ajbsent some officially coerced self-accusation, the Fifth Amendment privilege is not violated by even the most damning admissions.” United States v. Washington, 431 U. S. 181, 187 (1977) (emphasis added). The Miranda Court, however, presumed that interrogation in certain custodial circumstances4 is inherently coercive and held that statements made under those circumstances are inadmissible unless the suspect is specifically informed of his Miranda rights and freely decides to forgo those rights. The prophylactic Miranda warnings therefore are “not themselves rights protected by the Constitution but [are] instead measures to insure that the right against compulsory self-incrimination [is] protected.” Michigan v. Tucker, 417 U. S. 433, 444 (1974); see Edwards v. Arizona, 451 U. S. 477, 492 (1981) (Powell, J., concurring). Requiring Miranda warnings before custodial interrogation provides “practical reinforcement” for the Fifth Amendment right. Michigan v. Tucker, supra, at 444.

In this case we have before us no claim that respondent’s statements were actually compelled by police conduct which overcame his will to resist. See Beckwith v. United States, 425 U. S. 341, 347-348 (1976); Davis v. North Carolina, 384 U. S. 737 (1966). Thus the only issue before us is whether *655Officer Kraft was justified in failing to make available to respondent the procedural safeguards associated with the privilege against compulsory self-incrimination since Miranda.5

The New York Court of Appeals was undoubtedly correct in deciding that the facts of this case come within the ambit of the Miranda decision as we have subsequently interpreted it. We agree that respondent was in police custody because we have noted that “the ultimate inquiry is simply whether there is a ‘formal arrest or restraint on freedom of movement’ of the degree associated with a formal arrest,” California v. Beheler, 463 U. S. 1121, 1125 (1983) (per curiam), quoting Oregon v. Mathiason, 429 U. S. 492, 495 (1977) (per curiam). Here Quarles was surrounded by at least four police officers and was handcuffed when the questioning at issue took place. As the New York Court of Appeals observed, there was nothing to suggest that any of the officers were any longer concerned for their own physical safety. 58 N. Y. 2d, at 666, 444 N. E. 2d, at 985. The New York Court of Appeals’ majority declined to express an opinion as to whether there might be an exception to the Miranda rule if the police had been acting to protect the public, because the lower courts in New York had made no factual determination that the police had acted with that motive. Ibid.

We hold that on these facts there is a “public safety” exception to the requirement that Miranda warnings be given before a suspect’s answers may be admitted into evidence, *656and that the availability of that exception does not depend upon the motivation of the individual officers involved. In a kaleidoscopic situation such as the one confronting these officers, where spontaneity rather than adherence to a police manual is necessarily the order of the day, the application of the exception which we recognize today should not be made to depend on post hoc findings at a suppression hearing concerning the subjective motivation of the arresting officer.6 Undoubtedly most police officers, if placed in Officer Kraft’s position, would act out of a host of different, instinctive, and largely unverifiable motives — their own safety, the safety of others, and perhaps as well the desire to obtain incriminating evidence from the suspect.

Whatever the motivation of individual officers in such a situation, we do not believe that the doctrinal underpinnings of Miranda require that it be applied in all its rigor to a situation in which police officers ask questions reasonably prompted by a concern for the public safety. The Miranda decision was based in large part on this Court’s view that the warnings which it required police to give to suspects in custody would reduce the likelihood that the suspects would fall victim to constitutionally impermissible practices of police interrogation in the presumptively coercive environment of the station house. 384 U. S., at 455-458. The dissenters warned that the requirement of Miranda warnings would have the effect of decreasing the number of suspects who respond to police questioning. Id., at 504, 516-517 (Harlan, J., joined by Stewart and White, JJ., dissenting). The Miranda majority, however, apparently felt that whatever the *657cost to society in terms of fewer convictions of guilty suspects, that cost would simply have to be borne in the interest of enlarged protection for the Fifth Amendment privilege.

The police in this case, in the very act of apprehending a suspect, were confronted with the immediate necessity of ascertaining the whereabouts of a gun which they had every reason to believe the suspect had just removed from his empty holster and discarded in the supermarket. So long as the gun was concealed somewhere in the supermarket, with its actual whereabouts unknown, it obviously posed more than one danger to the public safety: an accomplice might make use of it, a customer or employee might later come upon it.

In such a situation, if the police are required to recite the familiar Miranda warnings before asking the whereabouts of the gun, suspects in Quarles’ position might well be deterred from responding. Procedural safeguards which deter a suspect from responding were deemed acceptable in Miranda in order to protect the Fifth Amendment privilege; when the primary social cost of those added protections is the possibility of fewer convictions, the Miranda majority was willing to bear that cost. Here, had Miranda warnings deterred Quarles from responding to Officer Kraft’s question about the whereabouts of the gun, the cost would have been something more than merely the failure to obtain evidence useful in convicting Quarles. Officer Kraft needed an answer to his question not simply to make his case against Quarles but to insure that further danger to the public did not result from the concealment of the gun in a public area.

We conclude that the need for answers to questions in a situation posing a threat to the public safety outweighs the need for the prophylactic rule protecting the Fifth Amendment’s privilege against self-incrimination. We decline to place officers such as Officer Kraft in the untenable position of having to consider, often in a matter of seconds, whether it best serves society for them to ask the necessary questions without the Miranda warnings and render whatever proba*658tive evidence they uncover inadmissible, or for them to give the warnings in order to preserve the admissibilty of evidence they might uncover but possibly damage or destroy their ability to obtain that evidence and neutralize the volatile situation confronting them.7 ■

In recognizing a narrow exception to the Miranda rule in this case, we acknowledge that to some degree we lessen the desirable clarity of that rule. At least in part in order to preserve its clarity, we have over the years refused to sanction attempts to expand our Miranda holding. See, e. g., Minnesota v. Murphy, 465 U. S. 420 (1984) (refusal to extend Miranda requirements to interviews with probation officers); Fare v. Michael C., 442 U. S. 707 (1979) (refusal to equate request to see a probation officer with request to see a lawyer for Miranda purposes); Beckwith v. United States, 425 U. S. 341 (1976) (refusal to extend Miranda requirements to questioning in noncustodial circumstances). As we have in other contexts, we recognize here the importance of a workable rule “to guide police officers, who have only limited time and expertise to reflect on and balance the social and individual interests involved in the specific circumstances they confront.” Dunaway v. New York, 442 U. S. 200, 213-214 (1979). But as we have pointed out, we believe that the exception which we recognize today lessens the necessity of that on-the-scene balancing process. The exception will not be difficult for police officers to apply because in each case it will be circumscribed by the exigency which justifies it. We think police officers can and will distinguish almost in*659stinctively between questions necessary to secure their own safety or the safety of the public and questions designed solely to elicit testimonial evidence from a suspect.

The facts of this case clearly demonstrate that distinction and an officer’s ability to recognize it. Officer Kraft asked only the question necessary to locate the missing gun before advising respondent of his rights. It was only after securing the loaded revolver and giving the warnings that he continued with investigatory questions about the ownership and place of purchase of the gun. The exception which we recognize today, far from complicating the thought processes and the on-the-scene judgments of police officers, will simply free them to follow their legitimate instincts when confronting situations presenting a danger to the public safety.8

We hold that the Court of Appeals in this case erred in excluding the statement, “the gun is over there,” and the gun because of the officer’s failure to read respondent his Miranda rights before attempting to locate the weapon. Ac*660cordingly we hold that it also erred in excluding the subsequent statements as illegal fruits of a Miranda violation.9 We therefore reverse and remand for further proceedings not inconsistent with this opinion.

It is so ordered.

Justice O’Connor,

concurring in the judgment in part and dissenting in part.

In Miranda v. Arizona, 384 U. S. 436 (1966), the Court held unconstitutional, because inherently compelled, the admission of statements derived from in-custody questioning not preceded by an explanation of the privilege against self-incrimination and the consequences of forgoing it. Today, the Court concludes that overriding considerations of public safety justify the admission of evidence — oral statements and a gun — secured without the benefit of such warnings. Ante, at 657-658. In so holding, the Court acknowledges that it is departing from prior precedent, see ante, at 653, and that it is “lessening] the desirable clarity of [the Miranda] rule,” ante, at 658. Were the Court writing from a clean slate, I could agree with its holding. But Miranda is now the law and, in my view, the Court has not provided sufficient justification for departing from it or for blurring its now clear strictures. Accordingly, I would require suppression of the initial statement taken from respondent in this case. On the other hand, nothing in Miranda or the privilege itself requires exclusion of nontestimonial evidence derived from informal custodial interrogation, and I therefore agree with the Court that admission of the gun in evidence is proper.1

*661Prior to Miranda, the privilege against self-incrimination had not been applied to an accused’s statements secured during custodial police interrogation. In these circumstances, the issue of admissibility turned, not on whether the accused had waived his privilege against self-incrimination, but on whether his statements were “voluntary” within the meaning of the Due Process Clause. See, e. g., Haynes v. Washington, 373 U. S. 503 (1963); Payne v. Arkansas, 356 U. S. 560 (1958); Chambers v. Florida, 309 U. S. 227 (1940); Brown v. Mississippi, 297 U. S. 278 (1936). Under this approach, the “totality of the circumstances” were assessed. If the interrogation was deemed unreasonable or shocking, or if the accused clearly did not have an opportunity to make a rational or intelligent choice, the statements received would be inadmissible.

The Miranda Court for the first time made the Self-Incrimination Clause applicable to responses induced by informal custodial police interrogation, thereby requiring suppression of many admissions that, under traditional due process principles, would have been admissible. More specifically, the Court held that

“the prosecution may not use statements, whether exculpatory or inculpatory, stemming from custodial interrogation of the defendant unless it demonstrates the use of *662procedural safeguards effective to secure the privilege against self-incrimination.” Miranda v. Arizona, 384 U. S., at 444.

Those safeguards included the now familiar Miranda warnings — namely, that the defendant must be informed

“that he has the right to remain silent, that anything he says can be used against him in a court of law, that he has the right to the presence of an attorney, and that if he cannot afford an attorney one will be appointed for him prior to any questioning if he so desires.” Id., at 479.

The defendant could waive these rights, but any waiver had to be made “knowingly and intelligently,” id., at 475, and the burden was placed on the prosecution to prove that such a waiver had voluntarily been made. Ibid. If the Miranda warnings were not properly administered or if no valid waiver could be shown, then all responses to interrogation made by the accused “while in custody ... or otherwise deprived of his freedom of action in any significant way” were to be presumed coerced and excluded from evidence at trial. Id., at 476, 479.

The Miranda Court itself considered objections akin to those raised by the Court today. In dissent, Justice White protested that the Miranda rules would “operate indiscriminately in all criminal cases, regardless of the severity of the crime or the circumstances involved.” Id., at 544. But the Miranda Court would not accept any suggestion that “society’s need for interrogation [could] outweig[h] the privilege.” To that Court, the privilege against self-incrimination was absolute and therefore could not be “abridged.” Id., at 479.

Since the time Miranda was decided, the Court has repeatedly refused to bend the literal terms of that decision. To be sure, the Court has been sensitive to the substantial burden *663the Miranda rules place on local law enforcement efforts, and consequently has refused to extend the decision or to increase its strictures on law enforcement agencies in almost any way. See, e. g., California v. Beheler, 463 U. S. 1121 (1983) (per curiam); Oregon v. Mathiason, 429 U. S. 492 (1977); Beckwith v. United States, 425 U. S. 341 (1976); Michigan v. Mosley, 423 U. S. 96 (1975); but cf.. Edwards v. Arizona, 451 U. S. 477 (1981). Similarly, where “statements taken in violation of the Miranda principles [have] not be[en] used to prove the prosecution’s case at trial,” the Court has allowed evidence derived from those statements to be admitted. Michigan v. Tucker, 417 U. S. 433, 445 (1974). But wherever an accused has been taken into “custody” and subjected to “interrogation” without warnings, the Court has consistently prohibited the use of his responses for prosecu-torial purposes at trial. See, e. g., Estelle v. Smith, 451 U. S. 454 (1981); Orozco v. Texas, 394 U. S. 324 (1969); Mathis v. United States, 391 U. S. 1 (1968); cf. Harris v. New York, 401 U. S. 222 (1971) (statements may be used for impeachment purposes). As a consequence, the “meaning of Miranda has become reasonably clear and law enforcement practices have adjusted to its strictures.” Rhode Island v. Innis, 446 U. S. 291, 304 (1980) (Burger, C. J., concurring); see generally Stephens, Flanders, & Cannon, Law Enforcement and the Supreme Court: Police Perceptions of the Miranda Requirements, 39 Tenn. L. Rev. 407 (1972).

In my view, a “public safety” exception unnecessarily blurs the edges of the clear line heretofore established and makes Miranda’s requirements more difficult to understand. In some cases, police will benefit because a reviewing court will find that an exigency excused their failure to administer the required warnings. But in other cases, police will suffer because, though they thought an exigency excused their noncompliance, a reviewing court will view the “objective” circumstances differently and require exclusion of admissions thereby obtained. The end result will be a finespun new *664doctrine on public safety exigencies incident to custodial interrogation, complete with the hair-splitting distinctions that currently plague our Fourth Amendment jurisprudence. “While the rigidity of the prophylactic rules was a principal weakness in the view of dissenters and critics outside the Court, . . . that rigidity [has also been called a] strength of the decision. It [has] afforded police and courts clear guidance on the manner in which to conduct a custodial investigation: if it was rigid, it was also precise. . . . [T]his core virtue of Miranda would be eviscerated if the prophylactic rules were freely [ignored] by . . . courts under the guise of [reinterpreting] Miranda . . . .” Fare v. Michael C., 439 U. S. 1310, 1314 (1978) (Rehnquist, J., in chambers on application for stay).

The justification the Court provides for upsetting the equilibrium that has finally been achieved — that police cannot and should not balance considerations of public safety against the individual’s interest in avoiding compulsory testimonial self-incrimination — really misses the critical question to be decided. See ante, at 657-658. Miranda has never been read to prohibit the police from asking questions to secure the public safety. Rather, the critical question Miranda addresses is who shall bear the cost of securing the public safety when such questions are asked and answered: the defendant or the State. Miranda, for better or worse, found the resolution of that question implicit in the prohibition against compulsory self-incrimination and placed the burden on the State. When police ask custodial questions without administering the required warnings, Miranda quite clearly requires that the answers received be presumed compelled and that they be excluded from evidence at trial. See Michigan v. Tucker, swpra, at 445, 447-448, 451, 452, and n. 26; Orozco v. Texas, supra, at 326.

The Court concedes, as it must, both that respondent was in “custody” and subject to “interrogation” and that his statement “the gun is over there” was compelled within the meaning of our precedent. See ante, at 654-655. In my view, *665since there is nothing about an exigency that makes custodial interrogation any less compelling, a principled application of Miranda requires that respondent’s statement be suppressed.

II

The court below assumed, without discussion, that the privilege against self-incrimination required that the gun derived from respondent’s statement also be suppressed, whether or not the State could independently link it to him.2 That conclusion was, in my view, incorrect.

A

Citizens in our society have a deeply rooted social obligation “to give whatever information they may have to aid in law enforcement.” Miranda v. Arizona, 384 U. S., at 478. *666Except where a recognized exception applies, “the criminal defendant no less than any other citizen is obliged to assist the authorities.” Roberts v. United States, 445 U. S. 552, 558 (1980). The privilege against compulsory self-incrimination is one recognized exception, but it is an exception nonetheless. Only the introduction of a defendant’s own testimony is proscribed by the Fifth Amendment’s mandate that no person “shall be compelled in any criminal case to be a witness against himself.” That mandate does not protect an accused from being compelled to surrender nontestimonial evidence against himself. See Fisher v. United States, 425 U. S. 391, 408 (1976).

The distinction between testimonial and nontestimonial evidence was explored in some detail in Schmerber v. California, 384 U. S. 757 (1966), a decision this Court handed down a week after deciding Miranda. The defendant in Schmerber had argued that the privilege against self-incrimination barred the State from compelling him to submit to a blood test, the results of which would be used to prove his guilt at trial. The State, on the other hand, had urged that the privilege prohibited it only from compelling the accused to make a formal testimonial statement against himself in an official legal proceeding. This Court rejected both positions. It favored an approach that protected the “accused only from being compelled to testify against himself, or otherwise provide the State with evidence of a testimonial or communicative nature.” 384 U. S., at 761. The blood tests were admissible because they were neither testimonial nor communicative in nature. Id., at 765.

In subsequent decisions, the Court relied on Schmerber in holding the privilege inapplicable to situations where the accused was compelled to stand in a lineup and utter words that allegedly had been spoken by the robber, see United States v. Wade, 388 U. S. 218, 221-223 (1967), to provide handwriting samples, see Gilbert v. California, 388 U. S. 263, 265-266 (1967), and to supply voice exemplars. See United States v. Dionisio, 410 U. S. 1, 5-7 (1973); see also United States v. *667 Mara, 410 U. S. 19, 21-22 (1973). “The distinction which . . . emerged [in these cases], often expressed in different ways, [was] that the privilege is a bar against compelling ‘communications’ or ‘testimony,’ but that compulsion which makes a suspect or accused the source of ‘real or physical evidence’ does not violate it.” Schmerber v. California, supra, at 764.

B

The gun respondent was compelled to supply is clearly evidence of the “real or physical” sort. What makes the question of its admissibility difficult is the fact that, in asking respondent to produce the gun, the police also “compelled” him, in the Miranda sense, to create an incriminating testimonial response. In other words, the case is problematic because police compelled respondent not only to provide the gun but also to admit that he knew where it was and that it was his.

It is settled that Miranda did not itself determine whether physical evidence obtained in this manner would be admissible. See Michigan v. Tucker, 417 U. S., at 445-446, 447, 452, and n. 26. But the Court in Schmerber, with Miranda fresh on its mind, did address the issue. In concluding that the privilege did not require suppression of compelled blood tests, the Court noted:

“This conclusion would not necessarily govern had the State tried to show that the accused had incriminated himself when told that he would have to be tested. Such incriminating evidence may be an unavoidable byproduct of the compulsion to take the test, especially for an individual who fears the extraction or opposes it on religious grounds. If it wishes to compel persons to submit to such attempts to discover evidence, the State may have to forgo the advantage of any testimonial products of administering the test — products which would fall within the privilege.” 384 U. S., at 765, and n. 9 (emphasis in original).

*668Thus, Schmerber resolved the dilemma by allowing admission of the nontestimonial, but not the testimonial, products of the State’s compulsion.

The Court has applied this bifurcated approach in its subsequent cases as well. For example, in United States v. Wade, 388 U. S. 218, 223 (1967), where admission of a lineup identification was approved, the Court emphasized that no question was presented as to the admissibility of anything said or done at the lineup. Likewise, in Michigan v. Tucker, where evidence derived from a technical Miranda violation was admitted, the Court noted that no statement taken without Miranda warnings was being admitted into evidence. See 417 U. S., at 445; cf. California v. Byers, 402 U. S. 424, 431-433 (1971) (opinion of Burger, C. J.). Thus, based on the distinction first articulated in Schmerber, “a strong analytical argument can be made for an intermediate rule whereby[,] although [the police] cannot require the suspect to speak by punishment or force, the nontestimonial [evidence derived from] speech that is [itself] excludable for failure to comply with the Miranda code could still be used.” H. Friendly, Benchmarks 280 (1967).

To be sure, admission of nontestimonial evidence secured through informal custodial interrogation will reduce the incentives to enforce the Miranda code. But that fact simply begs the question of how much enforcement is appropriate. There are some situations, as the Court’s struggle to accommodate a “public safety” exception demonstrates, in which the societal cost of administering the Miranda warnings is very high indeed .3 The Miranda decision quite practically does not express any societal interest in having those warn*669ings administered for their own sake. Rather, the warnings and waiver are only required to ensure that “testimony” used against the accused at trial is voluntarily given. Therefore, if the testimonial aspects of the accused’s custodial communications are suppressed, the failure to administer the Miranda warnings should cease to be of concern. Cf. Weatherford v. Bursey, 429 U. S. 545 (1977) (where interference with assistance of counsel has no effect on trial, no Sixth Amendment violation lies). The harm caused by failure to administer Miranda warnings relates only to admission of testimonial self-incriminations, and the suppression of such incrimina-tions should by itself produce the optimal enforcement of the Miranda rule.

C

There are, of course, decisions of this Court which suggest that the privilege against self-incrimination requires suppression not only of compelled statements but also of all evidence derived therefrom. See, e. g., Maness v. Meyers, 419 U. S. 449 (1975); Kastigar v. United States, 406 U. S. 441 (1972); McCarthy v. Arndstein, 266 U. S. 34 (1924); Counselman v. Hitchcock, 142 U. S. 547 (1892). In each of these cases, however, the Court was responding to the dilemma that confronts persons asserting their Fifth Amendment privilege to a court or other tribunal vested with the contempt power. In each instance, the tribunal can require witnesses to appear without any showing of probable cause to believe they have committed an offense or that they have relevant information to convey, and require the witnesses to testify even if they have formally and expressly asserted a privilege of silence. Individuals in this situation are faced with what Justice Goldberg once described as “the cruel trilemma of self-accusation, perjury, or contempt.” Murphy v. Waterfront Comm’n, 378 U. S. 52, 55 (1964). If the witness’ invocation of the privilege at trial is not to be defeated by the State’s refusal to let him remain silent at an earlier proceeding, the witness has to *670be protected “against the use of his compelled answers and evidence derived therefrom in any subsequent criminal case. ...” Lefkowitz v. Turley, 414 U. S. 70, 78 (1973).

By contrast, suspects subject to informal custodial police interrogation of the type involved in this case are not in the same position as witnesses required to appear before a court, grand jury, or other such formal tribunal. Where independent evidence leads police to a suspect, and probable cause justifies his arrest, the suspect cannot seriously urge that the police have somehow unfairly infringed on his right “to a private enclave where he may lead a private life.” Murphy v. Waterfront Comm’n, supra, at 55. Moreover, when a suspect interjects not the privilege itself but a post hoc complaint that the police failed to administer Miranda warnings, he invokes only an irrebuttable presumption that the interrogation was coercive. He does not show that a privilege was raised and that the police actually or overtly coerced him to provide testimony and other evidence to be used against him at trial. See Johnson v. New Jersey, 384 U. S. 719, 730 (1966). He could have remained silent and the interrogator could not have punished him for refusing to speak. Indeed, the accused is in the unique position of seeking the protection of the privilege without having timely asserted it. Cf. United States v. Kordel, 397 U. S. 1, 10 (1970) (failure to assert waives right to complain about testimonial compulsion). The person in police custody surely may sense that he is in “trouble,” Oregon v. Hass, 420 U. S. 714, 722 (1975), but he is in no position to protest that he faced the Hobson’s choice of self-accusation, perjury, or contempt. He therefore has a much less sympathetic case for obtaining the benefit of a broad suppression ruling. See Michigan v. Tucker, 417 U. S., at 444-451; cf. New Jersey v. Portash, 440 U. S. 450, 458-459 (1979).

Indeed, whatever case can be made for suppression evaporates when the statements themselves are not admitted, given the rationale of the Schmerber line of cases. Certainly *671interrogation which provides leads to other evidence does not offend the values underlying the Fifth Amendment privilege any more than the compulsory taking of blood samples, fingerprints, or voice exemplars, all of which may be compelled in an “attempt to discover evidence that might be used to prosecute [a defendant] for a criminal offense.” Schmerber v. California, 384 U. S., at 761. Use of a suspect’s answers “merely to find other evidence establishing his connection with the crime [simply] differs only by a shade from the permitted use for that purpose of his body or his blood.” H. Friendly, Benchmarks 280 (1967). The values underlying the privilege may justify exclusion of an unwarned person’s out-of-court statements, as perhaps they may justify exclusion of statements and derivative evidence compelled under the threat of contempt. But when the only evidence to be admitted is derivative evidence such as a gun — derived not from actual compulsion but from a statement taken in the absence of Miranda warnings — those values simply cannot require suppression, at least no more so than they would for other such nontestimonial evidence.4

*672On the other hand, if a suspect is subject to abusive police practices and actually or overtly compelled to speak, it is reasonable to infer both an unwillingness to speak and a perceptible assertion of the privilege. See Mincey v. Arizona, 437 U. S. 385, 396-402 (1978). Thus, when the Miranda violation consists of a deliberate and flagrant abuse of the accused’s constitutional rights, amounting to a denial of due process, application of a broader exclusionary rule is warranted. Of course, “a defendant raising [such] a coerced-confession claim . . . must first prevail in a voluntariness hearing before his confession and evidence derived from it [will] become inadmissible.” Kastigar v. United States, 406 U. S., at 462. By contrast, where the accused proves only that the police failed to administer the Miranda warnings, exclusion of the statement itself is all that will and should be required.5 Limitation of the Miranda prohibition to testimonial use of the statements themselves adequately serves the purposes of the privilege against self-incrimination.

I — I

In Miranda, the Court looked to the experience of countries like England, India, Scotland, and Ceylon in developing its code to regulate custodial interrogations. See Miranda *673v. Arizona, 384 U. S., at 486-489. Those countries had also adopted procedural rules to regulate the manner in which police secured confessions to be used against accused persons at trial. See Note, Developments in the Law — Confessions, 79 Harv. L. Rev. 935, 1090-1114 (1966). Confessions induced by trickery or physical abuse were never admissible at trial, and any confession secured without the required procedural safeguards could, in the courts’ discretion, be excluded on grounds of fairness or prejudice. See Gotlieb, Confirmation by Subsequent Facts, 72 L. Q. Rev. 209, 223-224 (1956). But nontestimonial evidence derived from all confessions “not blatantly coerced” was and still is admitted. Friendly, supra, at 282; see also Commissioners of Customs and Excise v. Harz, 1 All E. R. 177, 182 (1967); King v. Warickshall, 1 Leach 262, 168 Eng. Rep. 234 (K. B. 1783). Admission of nontestimonial evidence of this type is based on the very sensible view that procedural errors should not cause entire investigations and prosecutions to be lost. See Enker & Elsen, Counsel For the Suspect: Massiah v. United States and Escobedo v. Illinois, 49 Minn. L. Rev. 47, 80 (1964).

The learning of these countries was important to development of the initial Miranda rule. It therefore should be of equal importance in establishing the scope of the Miranda exclusionary rule today.6 I would apply that learning in this case and adhere to our precedents requiring that statements elicited in the absence of Miranda warnings be suppressed. But because nontestimonial evidence such as the gun should not be suppressed, I join in that part of the Court’s judgment *674that reverses and remands for further proceedings with the gun admissible as evidence against the accused.

Justice Marshall,

with whom Justice Brennan and Justice Stevens join,

dissenting.

The police in this case arrested a man suspected of possessing a firearm in violation of New York law. Once the suspect was in custody and found to be unarmed, the arresting officer initiated an interrogation. Without being advised of his right not to respond, the suspect incriminated himself by locating the gun. The majority concludes that the State may rely on this incriminating statement to convict the suspect of possessing a weapon. I disagree. The arresting officers had no legitimate reason to interrogate the suspect without advising him of his rights to remain silent and to obtain assistance of counsel. By finding on these facts justification for unconsented interrogation, the majority abandons the clear guidelines enunciated in Miranda v. Arizona, 384 U. S. 436 (1966), and condemns the American judiciary to a new era of post hoc inquiry into the propriety of custodial interrogations. More significantly and in direct conflict with this Court’s longstanding interpretation of the Fifth Amendment, the majority has endorsed the introduction of coerced self-incriminating statements in criminal prosecutions. I dissent.

I

Shortly after midnight on September 11, 1980, Officer Kraft and three other policemen entered an A & P supermarket in search of respondent Quarles, a rape suspect who was reportedly armed. After a brief chase, the officers cornered Quarles in the back of the store. As the other officers trained their guns on the suspect, Officer Kraft frisked Quarles and discovered an empty shoulder holster. Officer Kraft then handcuffed Quarles, and the other officers holstered their guns. With Quarles’ hands manacled behind *675his back and the other officers standing close by, Officer Kraft questioned Quarles: “Where is the gun?” Gesturing towards a stack of liquid-soap cartons a few feet away, Quarles responded: “The gun is over there.” Behind the cartons, the police found a loaded revolver. The State of New York subsequently failed to prosecute the alleged rape, and charged Quarles on a solitary count of criminal possession of a weapon in the third degree.1 As proof of the critical element of the offense, the State sought to introduce Quarles’ response to Officer Kraft’s question as well as the revolver found behind the cartons. The Criminal Term of the Supreme Court of the State of New York ordered both Quarles’ statement and the gun suppressed. The suppression order was affirmed first by the Appellate Division, 85 App. Div. 2d 936, 447 N. Y. S. 2d 84 (1981), and again by the New York Court of Appeals, 58 N. Y. 2d 664, 444 N. E. 2d 984 (1982) (mem.).

The majority’s entire analysis rests on the factual assumption that the public was at risk during Quarles’ interrogation. This assumption is completely in conflict with the facts as found by New York’s highest court. Before the interrogation began, Quarles had been “reduced to a condition of physical powerlessness.” Id., at 667, 444 N. E. 2d, at 986. Contrary to the majority’s speculations, ante, at 657, Quarles was not believed to have, nor did he in fact have, an accomplice to come to his rescue. When the questioning began, the arresting officers were sufficiently confident of their safety to put away their guns. As Officer Kraft acknowledged at the suppression hearing, “the situation was under control.” App. 35a. Based on Officer Kraft’s own testimony, the New York Court of Appeals found: “Nothing *676suggests that any of the officers was by that time concerned for his own physical safety.” 58 N. Y. 2d, at 666, 444 N. E. 2d, at 985. The Court of Appeals also determined that there was no evidence that the interrogation was prompted by the arresting officers’ concern for the public’s safety. Ibid.

The majority attempts to slip away from these unambiguous findings of New York’s highest court by proposing that danger be measured by objective facts rather than the subjective intentions of arresting officers. Ante, at 655-656. Though clever, this ploy was anticipated by the New York Court of Appeals: “[T]here is no evidence in the record before us that there were exigent circumstances posing a risk to the public safety . . . .” 58 N. Y. 2d, at 666, 444 N. E. 2d, at 985.

The New York court’s conclusion that neither Quarles nor his missing gun posed a threat to the public’s safety is amply supported by the evidence presented at the suppression hearing. Again contrary to the majority’s intimations, ante, at 657, no customers or employees were wandering about the store in danger of coming across Quarles’ discarded weapon. Although the supermarket was open to the public, Quarles’ arrest took place during the middle of the night when the store was apparently deserted except for the clerks at the check-out counter. The police could easily have cordoned off the store and searched for the missing gun. Had they done so, they would have found the gun forthwith. The police were well aware that Quarles had discarded his weapon somewhere near the scene of the arrest. As the State acknowledged before the New York Court of Appeals: “After Officer Kraft had handcuffed and frisked the defendant in the supermarket, he knew with a high degree of certainty that the defendant’s gun was within the immediate vicinity of the encounter. He undoubtedly would have searched for it in the carton a few feet away without the defendant having looked in that direction and saying that it was there.” Brief for Appellant in No. 2512/80 (N. Y. Ct. App.), p. 11 (emphasis added).

*677Earlier this Term, four Members of the majority joined an opinion stating: “[Questions of historical fact . . . must be determined, in the first instance, by state courts and deferred to, in the absence of ‘convincing evidence’ to the contrary, by the federal courts.” Rushen v. Spain, 464 U. S. 114, 120 (1983) (per curiam). In this case, there was convincing, indeed almost overwhelming, evidence to support the New York court’s conclusion that Quarles’ hidden weapon did not pose a risk either to the arresting officers or to the public. The majority ignores this evidence and sets aside the factual findings of the New York Court of Appeals. More cynical observers might well conclude that a state court’s findings of fact “deserv[e] a ‘high measure of deference,’” ibid, (quoting Sumner v. Mata, 455 U. S. 591, 598 (1982)), only when deference works against the interests of a criminal defendant.

I — I hH

The majority s treatment of the legal issues presented in this case is no less troubling than its abuse of the facts. Before today’s opinion, the Court had twice concluded that, under Miranda v. Arizona, 384 U. S. 436 (1966), police officers conducting custodial interrogations must advise suspects of their rights before any questions concerning the whereabouts of incriminating weapons can be asked. Rhode Island v. Innis, 446 U. S. 291, 298-302 (1980) (dicta); Orozco v. Texas, 394 U. S. 324 (1969) (holding).2 Now the majority departs from these cases and rules that police may withhold *678 Miranda warnings whenever custodial interrogations concern matters of public safety.3

The majority contends that the law, as it currently stands, places police officers in a dilemma whenever they interrogate a suspect who appears to know of some threat to the public’s safety. Ante, at 657. If the police interrogate the suspect without advising him of his rights, the suspect may reveal information that the authorities can use to defuse the threat, but the suspect’s statements will be inadmissible at trial. If, on the other hand, the police advise the suspect of his rights, the suspect may be deterred from responding to the police’s questions, and the risk to the public may continue unabated. According to the majority, the police must now choose between establishing the suspect’s guilt and safeguarding the public from danger.

The majority proposes to eliminate this dilemma by creating an exception to Miranda v. Arizona for custodial interrogations concerning matters of public safety. Ante, at 658-659. Under the majority’s exception, police would be permitted to interrogate suspects about such matters before the suspects have been advised of their constitutional rights. Without being “deterred” by the knowledge that they have a constitutional right not to respond, these suspects will be likely to answer the questions. Should the answers also be incriminating, the State would be free to introduce them as evidence in a criminal prosecution. Through this “narrow exception to the Miranda rule,” ante, at 658, the majority proposes to protect the public’s safety -without jeopardizing the prosecution of criminal defendants. I find in this reasoning an unwise and unprincipled departure from our Fifth Amendment precedents.

*679Before today’s opinion, the procedures established in Miranda v. Arizona had “the virtue of informing police and prosecutors with specificity as to what they may do in conducting custodial interrogation, and of informing courts under what circumstances statements obtained during such interrogation are not admissible.” Fare v. Michael C., 442 U. S. 707, 718 (1979); see Harryman v. Estelle, 616 F. 2d 870, 873-874 (CA5 1980) (en banc), cert. denied, 449 U. S. 860 (1980). In a chimerical quest for public safety, the majority has abandoned the rule that brought 18 years of doctrinal tranquility to the field of custodial interrogations. As the majority candidly concedes, ante, at 658, a public-safety exception destroys forever the clarity of Miranda for both law enforcement officers and members of the judiciary. The Court’s candor cannot mask what a serious loss the administration of justice has incurred.

This case is illustrative of the chaos the “public-safety” exception will unleash. The circumstances of Quarles’ arrest have never been in dispute. After the benefit of briefing and oral argument, the New York Court of Appeals, as previously noted, concluded that there was “no evidence in the record before us that there were exigent circumstances posing a risk to the public safety.” 58 N. Y. 2d, at 666, 444 N. E. 2d, at 985. Upon reviewing the same facts and hearing the same arguments, a majority of this Court has come to precisely the opposite conclusion: “So long as the gun was concealed somewhere in the supermarket, with its actual whereabouts unknown, it obviously posed more than one danger to the public safety. ...” Ante, at 657.

If after plenary review two appellate courts so fundamentally differ over the threat to public safety presented by the simple and uncontested facts of this case, one must seriously question how law enforcement officers will respond to the majority’s new rule in the confusion and haste of the real world. As The Chief Justice wrote in a similar context: “Few, if any, police officers are competent to make the kind *680of evaluation seemingly contemplated . . . .” Rhode Island v. Innis, 446 U. S., at 304 (concurring in judgment). Not only will police officers have to decide whether the objective facts of an arrest justify an unconsented custodial interrogation, they will also have to remember to interrupt the interrogation and read the suspect his Miranda warnings once the focus of the inquiry shifts from protecting the public’s safety to ascertaining the suspect’s guilt. Disagreements of the scope of the “public-safety” exception and mistakes in its application are inevitable.4

The end result, as Justice O’Connor predicts, will be “a finespun new doctrine on public safety exigencies incident to custodial interrogation, complete with the hair-splitting distinctions that currently plague our Fourth Amendment jurisprudence.” Ante, at 663-664. In the meantime, the courts will have to dedicate themselves to spinning this new web of doctrines, and the country’s law enforcement agencies will have to suffer patiently through the frustations of another period of constitutional uncertainty.

I — I J-H

Though unfortunate, the difficulty of administering the “public-safety” exception is not the most profound flaw in the majority’s decision. The majority has lost sight ofthe fact that Miranda v. Arizona and our earlier custodial-interrogation cases all implemented a constitutional privilege against self-incrimination. The rules established in these cases were designed to protect criminal defendants against prosecutions based on coerced self-incriminating statements. The majority today turns its back on these constitutional consider*681ations, and invites the government to prosecute through the use of what necessarily are coerced statements.

A

The majority’s error stems from a serious misunderstanding of Miranda v. Arizona and of the Fifth Amendment upon which that decision was based. The majority implies that Miranda consisted of no more than a judicial balancing act in which the benefits of “enlarged protection for the Fifth Amendment privilege” were weighed against “the cost to society in terms of fewer convictions of guilty suspects.” Ante, at 656-657. Supposedly because the scales tipped in favor of the privilege against self-incrimination, the Miranda Court erected a prophylactic barrier around statements made during custodial interrogations. The majority now proposes to return to the scales of social utility to calculate whether Miranda’s prophylactic rule remains cost-effective when threats to the public’s safety are added to the balance. The results of the majority’s “test” are announced with pseudo-scientific precision:

“We conclude that the need for answers to questions in a situation posing a threat to the public safety outweighs the need for the prophylactic rule protecting the Fifth Amendment’s privilege against self-incrimination.” Ante, at 657.

The majority misreads Miranda. Though the Miranda dissent prophesized dire consequences, see 384 U. S., at 504, 516-517 (Harlan, J., dissenting), the Miranda Court refused to allow such concerns to weaken the protections of the Constitution:

“A recurrent argument made in these cases is that society’s need for interrogation outweighs the privilege. This argument is not unfamiliar to this Court. The whole thrust of our foregoing discussion demonstrates that the Constitution has prescribed the rights of the individual when confronted with the power of government *682when it provided in the Fifth Amendment that an individual cannot be compelled to be a witness against himself. That right cannot be abridged.” Id., at 479 (citation omitted).

Whether society would be better off if the police warned suspects of their rights before beginning an interrogation or whether the advantages of giving such warnings would outweigh their costs did not inform the Miranda decision. On the contrary, the Miranda Court was concerned with the proscriptions of the Fifth Amendment, and, in particular, whether the Self-Incrimination Clause permits the government to prosecute individuals based on statements made in the course of custodial interrogations.

Miranda v. Arizona was the culmination of a century-long inquiry into how this Court should deal with confessions made during custodial interrogations. Long before Miranda, the Court had recognized that the Federal Government was prohibited from introducing at criminal trials compelled confessions, including confessions compelled in the course of custodial interrogations. In 1924, Justice Brandéis was reciting settled law when he wrote: “[A] confession obtained by compulsion must be excluded whatever may have been the character of the compulsion, and whether the compulsion was applied in a judicial proceeding or otherwise.” Wan v. United States, 266 U. S. 1, 14-15 (citing Bram v. United States, 168 U. S. 532 (1897)).

Prosecutors in state courts were subject to similar constitutional restrictions. Even before Malloy v. Hogan, 378 U. S. 1 (1964), formally applied the Self-Incrimination Clause of the Fifth Amendment to the States, the Due Process Clause constrained the States from extorting confessions from criminal defendants. Chambers v. Florida, 309 U. S. 227 (1940); Brown v. Mississippi, 297 U. S. 278 (1936). Indeed, by the time of Malloy, the constraints of the Due Process Clause were almost as stringent as the requirements of the Fifth Amendment itself. 378 U. S., at 6-7; see, e. g., Haynes v. Washington, 373 U. S. 503 (1963).

*683When Miranda reached this Court, it was undisputed that both the States and the Federal Government were constitutionally prohibited from prosecuting defendants with confessions coerced during custodial interrogations.5 As a theoretical matter, the law was clear. In practice, however, the courts found it exceedingly difficult to determine whether a given confession had been coerced. Difficulties of proof and subtleties of interrogation technique made it impossible in most cases for the judiciary to decide with confidence whether the defendant had voluntarily confessed his guilt or whether his testimony had been unconstitutionally compelled. Courts around the country were spending countless hours reviewing the facts of individual custodial interrogations. See Note, Developments in the Law — Confessions, 79 Harv. L. Rev. 935 (1966).

Miranda dealt with these practical problems. After a detailed examination of police practices and a review of its previous decisions in the area, the Court in Miranda determined that custodial interrogations are inherently coercive. The Court therefore created a constitutional presumption that statements made during custodial interrogations are compelled in violation of the Fifth Amendment and are thus inadmissible in criminal prosecutions. As a result of the Court’s decision in Miranda, a statement made during a custodial interrogation may be introduced as proof of a defendant’s guilt only if the prosecution demonstrates that the defendant knowingly and intelligently waived his constitutional rights before making the statement.6 The *684now-familiar Miranda warnings offer law enforcement authorities a clear, easily administered device for ensuring that criminal suspects understand their constitutional rights well enough to waive them and to engage in consensual custodial interrogation.

In fashioning its “public-safety” exception to Miranda, the majority makes no attempt to deal with the constitutional presumption established by that case. The majority does not argue that police questioning about issues of public safety is any less coercive than custodial interrogations into other matters. The majority’s only contention is that police officers could more easily protect the public if Miranda did not apply to custodial interrogations concerning the public’s safety.7 But Miranda was not a decision about public safety; it was a decision about coerced confessions. Without establishing that interrogations concerning the public’s safety are less likely to be coercive than other interrogations, the majority cannot endorse the “public-safety” exception and remain faithful to the logic of Miranda v. Arizona.

B

The majority’s avoidance of the issue of coercion may not have been inadvertent. It would strain credulity to contend *685that Officer Kraft’s questioning of respondent Quarles was not coercive.8 In the middle of the night and in the back of an empty supermarket, Quarles was surrounded by four armed police officers. His hands were handcuffed behind his back. The first words out of the mouth of the arresting officer were: “Where is the gun?” In the majority’s phrase, the situation was “kaleidoscopic.” Ante, at 656. Police and suspect were acting on instinct. Officer Kraft’s abrupt and pointed question pressured Quarles in precisely the way that the Miranda Court feared the custodial interrogations would coerce self-incriminating testimony.

That the application of the “public-safety” exception in this case entailed coercion is no happenstance. The majority’s ratio decidendi is that interrogating suspects about matters of public safety will be coercive. In its cost-benefit analysis, the Court’s strongest argument in favor of a “public-safety” exception to Miranda is that the police would be better able to protect the public’s safety if they were not always required to give suspects their Miranda warnings. The crux of this argument is that, by deliberately withholding Miranda warnings, the police can get information out of suspects who would refuse to respond to police questioning were they advised of their constitutional rights. The “public-safety” exception is efficacious precisely because it permits police officers to coerce criminal defendants into making involuntary statements.

Indeed, in the efficacy of the “public-safety” exception lies a fundamental and constitutional defect. Until today, this Court could truthfully state that the Fifth Amendment is given “broad scope” “[w]here there has been genuine compul*686sion of testimony.” Michigan v. Tucker, 417 U. S. 433, 440 (1974). Coerced confessions were simply inadmissible in criminal prosecutions. The “public-safety” exception departs from this principle by expressly inviting police officers to coerce defendants into making incriminating statements, and then permitting prosecutors to introduce those statements at trial. Though the majority’s opinion is cloaked in the beguiling language of utilitarianism, the Court has sanctioned sub silentio criminal prosecutions based on compelled self-incriminating statements. I find this result in direct conflict with the Fifth Amendment’s dictate that “[n]o person . . . shall be compelled in any criminal case to be a witness against himself.”

The irony of the majority’s decision is that the public’s safety can be perfectly well protected without abridging the Fifth Amendment. If a bomb is about to explode or the public is otherwise imminently imperiled, the police are free to interrogate suspects without advising them of their constitutional rights. Such unconsented questioning may take place not only when police officers act on instinct but also when higher faculties lead them to believe that advising a suspect of his constitutional rights might decrease the likelihood that the suspect would reveal life-saving information. If trickery is necessary to protect the public, then the police may trick a suspect into confessing. While the Fourteenth Amendment sets limits on such behavior, nothing in the Fifth Amendment or our decision in Miranda v. Arizona proscribes this sort of emergency questioning. All the Fifth Amendment forbids is the introduction of coerced statements at trial. Cf. Weatherford v. Bursey, 429 U. S. 545 (1977) (Sixth Amendment violated only if trial affected).

To a limited degree, the majority is correct that there is a cost associated with the Fifth Amendment’s ban on introducing coerced self-incriminating statements at trial. Without a “public-safety” exception, there would be occasions when a defendant incriminated himself by revealing a threat to the *687public, and the State was unable to prosecute because the defendant retracted his statement after consulting with counsel and the police cannot find independent proof of guilt. Such occasions would not, however, be common. The prosecution does not always lose the use of incriminating information revealed in these situations. After consulting with counsel, a suspect may well volunteer to repeat his statement in hopes of gaining a favorable plea bargain or more lenient sentence. The majority thus overstates its case when it suggests that a police officer must necessarily choose between public safety and admissibility.9

But however frequently or infrequently such cases arise, their regularity is irrelevant. The Fifth Amendment prohibits compelled self-incrimination.10 As the Court has explained on numerous occasions, this prohibition is the mainstay of our adversarial system of criminal justice. Not only does it protect us against the inherent unreliability of compelled testimony, but it also ensures that criminal investigations will be conducted with integrity and that the judiciary will avoid the taint of official lawlessness. See Murphy *688v. Waterfront Comm’n, 378 U. S. 52, 55 (1964). The policies underlying the Fifth Amendment’s privilege against self-incrimination are not diminished simply because testimony is compelled to protect the public’s safety. The majority should not be permitted to elude the Amendment’s absolute prohibition simply by calculating special costs that arise when the public’s safety is at issue. Indeed, were constitutional adjudication always conducted in such an ad hoc manner, the Bill of Rights would be a most unreliable protector of individual liberties.

IV

Having determined that the Fifth Amendment renders inadmissible Quarles’ response to Officer Kraft’s questioning, I have no doubt that our precedents require that the gun discovered as a direct result of Quarles’ statement must be presumed inadmissible as well. The gun was the direct product of a coercive custodial interrogation. In Silver-thome Lumber Co. v. United States, 251 U. S. 385 (1920), and Wong Sun v. United States, 371 U. S. 471 (1963), this Court held that the Government may not introduce incriminating evidence derived from an illegally obtained source. This Court recently explained the extent of the Wong Sun rule:

“Although Silverthome and Wong Sun involved violations of the Fourth Amendment, the ‘fruit of the poisonous tree’ doctrine has not been limited to cases in which there has been a Fourth Amendment violation. The Court has applied the doctrine where the violations were of the Sixth Amendment, see United States v. Wade, 388 U. S. 218 (1967), as well as of the Fifth Amendment.” Nix v. Williams, ante, at 442 (footnote omitted).

Accord, United States v. Crews, 445 U. S. 463, 470 (1980).11 When they ruled on the issue, the New York courts were *689entirely correct in deciding that Quarles’ gun was the tainted fruit of a nonconsensual interrogation and therefore was inadmissible under our precedents.

However, since the New York Court of Appeals issued its opinion, the scope of the Wong Sun doctrine has changed. In Nix v. Williams, supra, this Court construed Wong Sun to permit the introduction into evidence of constitutionally tainted “fruits” that inevitably would have been discovered by the government. In its briefs before this Court and before the New York courts, petitioner has argued that the “inevitable-discovery” rule, if applied to this case, would permit the admission of Quarles’ gun. Although I have not joined the Court’s opinion in Nix, and although I am not wholly persuaded that New York law would permit the application of the “inevitable-discovery” rule to this case,12 *690I believe that the proper disposition of the matter is to vacate the order of the New York Court of Appeals to the extent that it suppressed Quarles’ gun and remand the matter to the New York Court of Appeals for further consideration in light of Nix v. Williams.

Accordingly, I would affirm the order of the Court of Appeals to the extent that it found Quarles’ incriminating statement inadmissible under the Fifth Amendment, would vacate the order to the extent that it suppressed Quarles’ gun, and would remand the matter for reconsideration in light of Nix v. Williams.