Justice Scalia delivered the opinion of the Court.

                . . . On August 5, 1999, Kenneth Lee was stabbed at his apartment.  Police arrested petitioner [Michael Crawford] later that night.  After giving petitioner and his wife [Sylvia] Miranda warnings, detectives interrogated each of them twice.  Petitioner eventually confessed that he and Sylvia had gone in search of Lee because he was upset over an earlier incident in which Lee had tried to rape her.  The two had found Lee at his apartment, and a fight ensued . . . .  Sylvia generally corroborated petitioner’s story abut the events leading up to the fight, but her account of the fight itself was arguably different—particularly with respect to whether Lee had drawn a weapon before petitioner assaulted him . . . .

                The State charged petitioner with assault and attempted murder.  At trial, he claimed self-defense.  Sylvia did not testify because of the state marital privilege, which generally bars a spouse from testifying without the other spouse’s consent.  See Wash. Rev. Code §5.60.060(1) (1994).  In Washington, this privilege does not extend to a spouse’s out-of-court statements admissible under a hearsay exception, see State v. Burden, [841 P.2d 758, 761 (Wash. 1992)], so the State sought to introduce Sylvia’s tape-recorded statements to the police as evidence that the stabbing was not in self-defense.  Noting that Sylvia had admitted she led petitioner to Lee’s apartment and thus had facilitated the assault, the State invoked the hearsay exception for statements against penal interest, Wash. Rule Evid. 804(b)(3) (2003). . . .

                According to our description of [the confrontation] right in Ohio v. Roberts, 448 U.S. 56 (1980), it does not bar admission of an unavailable witness’s statement against a criminal defendant if the statement bears “adequate ‘indicia of reliability.’”  To meet that test, evidence must either fall within a “firmly rooted hearsay exception” or bear “particularized guarantees of trustworthiness.”  The trial court here admitted the statement on the latter ground, offering several reasons why it was trustworthy:  Sylvia was not shifting blame but rather corroborating her husband’s story that he acted in self-defense or “justified reprisal”; she had direct knowledge as an eyewitness; she was describing recent events; and she was being questioned by a “neutral” law enforcement officer.  The prosecution played the tape for the jury and relied on it in closing, arguing that it was “damning evidence” that “completely refutes [petitioner’s] claim of self-defense.”  The jury convicted petitioner of assault.

                The Washington Court of Appeals reversed. . . .  The Washington Supreme Court reinstated the conviction, unanimously concluding that, although Sylvia’s statement did not fall under a firmly rooted hearsay exception, it bore guarantees of trustworthiness:  “‘[W]hen a codefendant’s confession is virtually identical [to, i.e., interlocks with,] that of a defendant, it may be deemed reliable.’”. . . 1 . . .

                The right to confront one’s accusers is a concept that dates back to Roman times. . . .  The founding generation’s immediate source of the concept, however, was the common law.  English common law has long differed from continental civil law in regard to the manner in which witnesses give testimony in criminal trials.  The common-law tradition is one of live testimony in court subject to adversarial testing, while the civil law condones examination in private by judicial officers.  See 3 W. Blackstone, Commentaries on the Laws of England 373-374 (1768).

                Nonetheless, England at times adopted elements of the civil-law practice. . . .  The most notorious instances of civil-law examination occurred in the great political trials of the 16th and 17th centuries.  One such was the 1603 trial of Sir Walter Raleigh for treason. Lord Cobham, Raleigh’s alleged accomplice, had implicated him in an examination before the Privy Council and in a letter.  At Raleigh’s trial, these were read to the jury. Raleigh argued that Cobham had lied to save himself:  “Cobham is absolutely in the King’s mercy; to excuse me cannot avail him; by accusing me he may hope for favour.”  1 D. Jardine, Criminal Trials 435 (1832).  Suspecting that Cobham would recant, Raleigh demanded that the judges call him to appear, arguing that “[t]he Proof of the Common Law is by witness and jury:  let Cobham be here, let him speak it.  Call my accuser before my face . . . .”  [Raleigh’s Case, 2 How. St. Tr. 1, 15-16 (1603)].  The judges refused, and, despite Raleigh’s protestations that he was being tried “by the Spanish Inquisition,” the jury convicted, and Raleigh was sentenced to death.

                One of Raleigh’s trial judges later lamented that “‘the justice of England has never been so degraded and injured as by the condemnation of Sir Walter Raleigh.’” 1 Jardine, supra, at 520.  Through a series of statutory and judicial reforms, English law developed a right of confrontation that limited these abuses.  For example, treason statutes required witnesses to confront the accused “face to face” at his arraignment.  Courts, meanwhile, developed relatively strict rules of unavailability, admitting examinations only if the witness was demonstrably unable to testify in person. . . . 

                One recurring question was whether the admissibility of an unavailable witness’s pretrial examination depended on whether the defendant had had an opportunity to cross-examine him.  In 1696, the Court of King’s Bench answered this question in the affirmative, in the widely reported misdemeanor libel case of King v. Paine, 5 Mod. 163, 87 Eng. Rep. 584. The court ruled that, even though a witness was dead, his examination was not admissible where “the defendant not being present when [it was] taken before the mayor . . . had lost the benefit of a cross-examination.” . . . [B]y 1791 (the year the Sixth Amendment was ratified), courts were applying the cross-examination rule even to examinations by justices of the peace in felony cases. . . .

                Controversial examination practices were also used in the Colonies. . . .  [Partly as a result, many] declarations of rights adopted around the time of the Revolution guaranteed a right of confrontation. . . .   The proposed Federal Constitution, however, did not.  At the Massachusetts ratifying convention, Abraham Holmes objected to this omission precisely on the ground that it would lead to civil-law practices. . . .  Similarly, a prominent Antifederalist writing under the pseudonym Federal Farmer criticized the use of “written evidence” . . . .  The First Congress responded by including the Confrontation Clause in the proposal that became the Sixth Amendment. . . .

                This history supports two inferences about the meaning of the Sixth Amendment.

                First, the principal evil at which the Confrontation Clause was directed was the civil-law mode of criminal procedure, and particularly its use of ex parte examinations as evidence against the accused. . . .  The Sixth Amendment must be interpreted with this focus in mind.

                Accordingly, we once again reject the view that the Confrontation Clause applies of its own force only to in-court testimony, and that its application to out-of-court statements introduced at trial depends upon “the law of Evidence for the time being.” 3 Wigmore §  1397, at 101 . . . .  Leaving the regulation of out-of-court statements to the law of evidence would render the Confrontation Clause powerless to prevent even the most flagrant inquisitorial practices.  Raleigh was, after all, perfectly free to confront those who read Cobham’s confession in court.

                This focus also suggests that not all hearsay implicates the Sixth Amendment’s core concerns.  An off-hand, overheard remark might be unreliable evidence and thus a good candidate for exclusion under hearsay rules, but it bears little resemblance to the civil-law abuses the Confrontation Clause targeted.  On the other hand, ex parte examinations might sometimes be admissible under modern hearsay rules, but the Framers certainly would not have condoned them.

                The text of the Confrontation Clause reflects this focus.  It applies to “witnesses” against the accused—in other words, those who “bear testimony.” 1 N. Webster, An American Dictionary of the English Language (1828).  “Testimony,” in turn, is typically “[a] solemn declaration or affirmation made for the purpose of establishing or proving some fact.” Ibid.  An accuser who makes a formal statement to government officers bears testimony in a sense that a person who makes a casual remark to an acquaintance does not.  The constitutional text, like the history underlying the common-law right of confrontation, thus reflects an especially acute concern with a specific type of out-of-court statement.

                Various formulations of this core class of “testimonial” statements exist: “ex parte in-court testimony or its functional equivalent—that is, material such as affidavits, custodial examinations, prior testimony that the defendant was unable to cross-examine, or similar pretrial statements that declarants would reasonably expect to be used prosecutorially,” Brief for Petitioner 23; “extrajudicial statements . . . contained in formalized testimonial materials, such as affidavits, depositions, prior testimony, or confessions,” White v. Illinois, 502 U.S. 346, 365 (1992) (Thomas, J., joined by Scalia, J., concurring in part and concurring in judgment); “statements that were made under circumstances which would lead an objective witness reasonably to believe that the statement would be available for use at a later trial,” Brief for National Association of Criminal Defense Lawyers et al. as Amici Curiae 3. These formulations all share a common nucleus and then define the Clause’s coverage at various levels of abstraction around it. Regardless of the precise articulation, some statements qualify under any definition—for example, ex parte testimony at a preliminary hearing.

                Statements taken by police officers in the course of interrogations are also testimonial under even a narrow standard.  Police interrogations bear a striking resemblance to examinations by justices of the peace in England.  The statements are not sworn testimony, but the absence of oath was not dispositive. Cobham’s examination was unsworn, yet Raleigh’s trial has long been thought a paradigmatic confrontation violation. . . . 3  . . .  In sum, even if the Sixth Amendment is not solely concerned with testimonial hearsay, that is its primary object, and interrogations by law enforcement officers fall squarely within that class.

                The historical record also supports a second proposition:  that the Framers would not have allowed admission of testimonial statements of a witness who did not appear at trial unless he was unavailable to testify, and the defendant had had a prior opportunity for cross-examination.  The text of the Sixth Amendment does not suggest any open-ended exceptions from the confrontation requirement to be developed by the courts.  Rather, the “right . . . to be confronted with the witnesses against him” is most naturally read as a reference to the right of confrontation at common law, admitting only those exceptions established at the time of the founding. . . . [T]he common law in 1791 conditioned admissibility of an absent witness’s examination on unavailability and a prior opportunity to cross-examine.  The Sixth Amendment therefore incorporates those limitations. . . .

                This is not to deny, as The Chief Justice notes, that “[t]here were always exceptions to the general rule of exclusion” of hearsay evidence.  Several had become well established by 1791.  But there is scant evidence that exceptions were invoked to admit testimonial statements against the accused in a criminal case.6  Most of the hearsay exceptions covered statements that by their nature were not testimonial—for example, business records or statements in furtherance of a conspiracy.  We do not infer from these that the Framers thought exceptions would apply even to prior testimony. . . . 7

                Our case law has been largely consistent with these two principles . . .  8 . . .  [but] the same cannot be said of our rationales.  Roberts conditions the admissibility of all hearsay evidence on whether it falls under a “firmly rooted hearsay exception” or bears “particularized guarantees of trustworthiness.”  This test departs from the historical principles identified above in two respects.  First, it is too broad:  It applies the same mode of analysis whether or not the hearsay consists of ex parte testimony.  This often results in close constitutional scrutiny in cases that are far removed from the core concerns of the Clause.  At the same time, however, the test is too narrow:  It admits statements that do consist of ex parte testimony upon a mere finding of reliability.  This malleable standard often fails to protect against paradigmatic confrontation violations.

                Members of this Court and academics have suggested that we revise our doctrine to reflect more accurately the original understanding of the Clause. . . .  They offer two proposals:  First, that we apply the Confrontation Clause only to testimonial statements, leaving the remainder to regulation by hearsay law—thus eliminating the overbreadth referred to above.  Second, that we impose an absolute bar to statements that are testimonial, absent a prior opportunity to cross-examine—thus eliminating the excessive narrowness referred to above.

                In White, we considered the first proposal and rejected it.  Although our analysis in this case casts doubt on that holding, we need not definitively resolve whether it survives our decision today, because Sylvia Crawford’s statement is testimonial under any definition. This case does, however, squarely implicate the second proposal.

                Where testimonial statements are involved, we do not think the Framers meant to leave the Sixth Amendment’s protection to the vagaries of the rules of evidence, much less to amorphous notions of “reliability.”  Certainly none of the authorities discussed above acknowledges any general reliability exception to the common-law rule.  Admitting statements deemed reliable by a judge is fundamentally at odds with the right of confrontation.  To be sure, the Clause’s ultimate goal is to ensure reliability of evidence, but it is a procedural rather than a substantive guarantee.  It commands, not that evidence be reliable, but that reliability be assessed in a particular manner:  by testing in the crucible of cross-examination.  The Clause thus reflects a judgment, not only about the desirability of reliable evidence (a point on which there could be little dissent), but about how reliability can best be determined.  Cf. 3 Blackstone, Commentaries, at 373 (“This open examination of witnesses . . . is much more conducive to the clearing up of truth”); M. Hale, History and Analysis of the Common Law of England 258 (1713) (adversarial testing “beats and bolts out the Truth much better”).

                The Roberts test allows a jury to hear evidence, untested by the adversary process, based on a mere judicial determination of reliability.  It thus replaces the constitutionally prescribed method of assessing reliability with a wholly foreign one.  In this respect, it is very different from exceptions to the Confrontation Clause that make no claim to be a surrogate means of assessing reliability.  For example, the rule of forfeiture by wrongdoing (which we accept) extinguishes confrontation claims on essentially equitable grounds; it does not purport to be an alternative means of determining reliability. See Reynolds v. United States, 98 U.S. 145, 158-159 (1879).

                The Raleigh trial itself involved the very sorts of reliability determinations that Roberts authorizes.  In the face of Raleigh’s repeated demands for confrontation, the prosecution responded with many of the arguments a court applying Roberts might invoke today:  that Cobham’s statements were self-inculpatory, that they were not made in the heat of passion, and that they were not “extracted from [him] upon any hopes or promise of Pardon.”  It is not plausible that the Framers’ only objection to the trial was that Raleigh’s judges did not properly weigh these factors before sentencing him to death. Rather, the problem was that the judges refused to allow Raleigh to confront Cobham in court, where he could cross-examine him and try to expose his accusation as a lie.

                Dispensing with confrontation because testimony is obviously reliable is akin to dispensing with jury trial because a defendant is obviously guilty.  This is not what the Sixth Amendment prescribes.

                The legacy of Roberts in other courts vindicates the Framers’ wisdom in rejecting a general reliability exception. The framework is so unpredictable that it fails to provide meaningful protection from even core confrontation violations.

                Reliability is an amorphous, if not entirely subjective, concept. . . .  Whether a statement is deemed reliable depends heavily on which factors the judge considers and how much weight he accords each of them.  Some courts wind up attaching the same significance to opposite facts.  For example, the Colorado Supreme Court held a statement more reliable because its inculpation of the defendant was “detailed,” [People v. Farrell, 34 P. 3d 401, 406-407 (2001)], while the Fourth Circuit found a statement more reliable because the portion implicating another was “fleeting,” United States v. Photogrammetric Data Servs., Inc., 259 F.3d 229, 245 (2001).  The Virginia Court of Appeals found a statement more reliable because the witness was in custody and charged with a crime (thus making the statement more obviously against her penal interest), see Nowlin v. Commonwealth, [579 S.E.2d 367, 371-372 (2003)], while the Wisconsin Court of Appeals found a statement more reliable because the witness was not in custody and not a suspect, see State v. Bintz, [2002 WI App. 204]. Finally, the Colorado Supreme Court in one case found a statement more reliable because it was given “immediately after” the events at issue, Farrell, supra, at 407, while that same court, in another case, found a statement more reliable because two years had elapsed, Stevens v. People, 29 P.3d 305, 316 (2001).

                The unpardonable vice of the Roberts test, however, is not its unpredictability, but its demonstrated capacity to admit core testimonial statements that the Confrontation Clause plainly meant to exclude. . . .  To add insult to injury, some of the courts that admit untested testimonial statements find reliability in the very factors that make the statements testimonial.  As noted earlier, one court relied on the fact that the witness’s statement was made to police while in custody on pending charges—the theory being that this made the statement more clearly against penal interest and thus more reliable. Other courts routinely rely on the fact that a prior statement is given under oath in judicial proceedings. E.g., [United States v. Gallego, 191 F.3d 156, 168 (CA2 1999)] (plea allocution); [United States v. Papajohn, 212 F.3d 1112, 1118-1120 (CA8 2000)] (grand jury testimony).  That inculpating statements are given in a testimonial setting is not an antidote to the confrontation problem, but rather the trigger that makes the Clause’s demands most urgent.  It is not enough to point out that most of the usual safeguards of the adversary process attend the statement, when the single safeguard missing is the one the Confrontation Clause demands.

                Roberts’ failings were on full display in the proceedings below.  Sylvia Crawford made her statement while in police custody, herself a potential suspect in the case.  Indeed, she had been told that whether she would be released “depend[ed] on how the investigation continues.”  In response to often leading questions from police detectives, she implicated her husband in Lee’s stabbing and at least arguably undermined his self-defense claim.  Despite all this, the trial court admitted her statement, listing several reasons why it was reliable.  In its opinion reversing, the Court of Appeals listed several other reasons why the statement was not reliable.  Finally, the State Supreme Court relied exclusively on the interlocking character of the statement and disregarded every other factor the lower courts had considered.  The case is thus a self-contained demonstration of Roberts’ unpredictable and inconsistent application.

                Each of the courts also made assumptions that cross-examination might well have undermined. The trial court, for example, stated that Sylvia Crawford’s statement was reliable because she was an eyewitness with direct knowledge of the events.  But Sylvia at one point told the police that she had “shut [her] eyes and . . . didn’t really watch” part of the fight, and that she was “in shock.” The trial court also buttressed its reliability finding by claiming that Sylvia was “being questioned by law enforcement, and, thus, the [questioner] is . . . neutral to her and not someone who would be inclined to advance her interests and shade her version of the truth unfavorably toward the defendant.” The Framers would be astounded to learn that ex parte testimony could be admitted against a criminal defendant because it was elicited by “neutral” government officers.  But even if the court’s assessment of the officer’s motives was accurate, it says nothing about Sylvia’s perception of her situation.  Only cross-examination could reveal that.

                The State Supreme Court gave dispositive weight to the interlocking nature of the two statements—that they were both ambiguous as to when and whether Lee had a weapon. The court’s claim that the two statements were equally ambiguous is hard to accept. Petitioner’s statement is ambiguous only in the sense that he had lingering doubts about his recollection:  “A. I coulda swore I seen him goin’ for somethin’ before, right before everything happened . . . .  [B]ut I’m not positive.”  Sylvia’s statement, on the other hand, is truly inscrutable, since the key timing detail was simply assumed in the leading question she was asked: “Q. Did Kenny do anything to fight back from this assault?”  Moreover, Sylvia specifically said Lee had nothing in his hands after he was stabbed, while petitioner was not asked about that.

                The prosecutor obviously did not share the court’s view that Sylvia’s statement was ambiguous —he called it “damning evidence” that “completely refutes [petitioner’s] claim of self-defense.”  We have no way of knowing whether the jury agreed with the prosecutor or the court.  Far from obviating the need for cross-examination, the “interlocking” ambiguity of the two statements made it all the more imperative that they be tested to tease out the truth.

                We readily concede that we could resolve this case by simply reweighing the “reliability factors” under Roberts and finding that Sylvia Crawford’s statement falls short.  But we view this as one of those rare cases in which the result below is so improbable that it reveals a fundamental failure on our part to interpret the Constitution in a way that secures its intended constraint on judicial discretion.  Moreover, to reverse the Washington Supreme Court’s decision after conducting our own reliability analysis would perpetuate, not avoid, what the Sixth Amendment condemns. The Constitution prescribes a procedure for determining the reliability of testimony in criminal trials, and we, no less than the state courts, lack authority to replace it with one of our own devising. . . .  Vague standards are manipulable, and, while that might be a small concern in run-of-the-mill assault prosecutions like this one, the Framers had an eye toward politically charged cases like Raleigh’s—great state trials where the impartiality of even those at the highest levels of the judiciary might not be so clear.  It is difficult to imagine Roberts’ providing any meaningful protection in those circumstances.

                Where nontestimonial hearsay is at issue, it is wholly consistent with the Framers’ design to afford the States flexibility in their development of hearsay law—as does Roberts, and as would an approach that exempted such statements from Confrontation Clause scrutiny altogether.  Where testimonial evidence is at issue, however, the Sixth Amendment demands what the common law required:  unavailability and a prior opportunity for cross-examination.  We leave for another day any effort to spell out a comprehensive definition of “testimonial.”10  Whatever else the term covers, it applies at a minimum to prior testimony at a preliminary hearing, before a grand jury, or at a former trial; and to police interrogations.  These are the modern practices with closest kinship to the abuses at which the Confrontation Clause was directed. . . .  The judgment of the Washington Supreme Court is reversed . . . . 

                Chief Justice Rehnquist, with whom Justice O’Connor joins, concurring in the judgment.

                I dissent from the Court’s decision to overrule Ohio v. Roberts, 448 U.S. 56 (1980). . . .  Under the common law, although the courts were far from consistent, out-of-court statements made by someone other than the accused and not taken under oath, unlike ex parte depositions or affidavits, were generally not considered substantive evidence upon which a conviction could be based. . . .  As far as I can tell, unsworn testimonial statements were treated no differently at common law than were nontestimonial statements, and it seems to me any classification of statements as testimonial beyond that of sworn affidavits and depositions will be somewhat arbitrary, merely a proxy for what the Framers might have intended had such evidence been liberally admitted as substantive evidence like it is today. . . .  I see little value in trading our precedent for an imprecise approximation at this late date. . . .

                Between 1700 and 1800 the rules regarding the admissibility of out-of-court statements were still being developed.  There were always exceptions to the general rule of exclusion, and it is not clear to me that the Framers categorically wanted to eliminate further ones. . . .  It is an odd conclusion indeed to think that the Framers created a cut-and-dried rule with respect to the admissibility of testimonial statements when the law during their own time was not fully settled. . . .

                Exceptions to confrontation have always been derived from the experience that some out-of-court statements are just as reliable as cross-examined in-court testimony due to the circumstances under which they were made. . . .  That a statement might be testimonial does nothing to undermine the wisdom of one of these exceptions. . . .

                 The Court grandly declares that “[w]e leave for another day any effort to spell out a comprehensive definition of ‘testimonial.’”  But the thousands of federal prosecutors and the tens of thousands of state prosecutors need answers as to what beyond the specific kinds of “testimony” the Court lists is covered by the new rule.  They need them now, not months or years from now.  Rules of criminal evidence are applied every day in courts throughout the country, and parties should not be left in the dark in this manner. . . .

                In Idaho v. Wright, 497 U.S. 805 (1990), we held that an out-of-court statement was not admissible simply because the truthfulness of that statement was corroborated by other evidence at trial.  As the Court notes, the Supreme Court of Washington gave decisive weight to the “interlocking nature of the two statements.”  No re-weighing of the “reliability factors,” which is hypothesized by the Court, is required to reverse the judgment here.  A citation to Idaho v. Wright would suffice. . . .  I believe that this would be a far preferable course for the Court to take here.

________________________