delivered the opinion of the Court, except as to Part IV and footnote 6.†
In Melendez-Diaz v. Massachusetts, 557 U. S. 305 (2009), this Court held that a forensic laboratory report stating that a suspect substance was cocaine ranked as testimonial for purposes of the Sixth Amendment’s Confrontation Clause. The report had been created specifically to serve as evidence in a criminal proceeding. Absent stipulation, the Court ruled, the prosecution may not introduce such a report without offering a live witness competent to testify to the truth of the statements made in the report.
In the case before us, petitioner Donald Bullcoming was arrested on charges of driving while intoxicated (DWI). Principal evidence against Bullcoming was a forensic laboratory report certifying that Bullcoming’s blood-alcohol concentration was well above the threshold for aggravated DWI. At trial, the prosecution did not call as a witness the analyst who signed the certification. Instead, the State called another analyst who was familiar with the laboratory’s testing procedures, but had neither participated in nor observed the test on Bullcoming’s blood sample. The New Mexico Supreme Court determined that, although the blood-alcohol analysis was “testimonial,” the Confrontation Clause did not require the certifying analyst’s in-court testimony. Instead, *652New Mexico’s high court held, live testimony of another analyst satisfied the constitutional requirements.
The question presented is whether the Confrontation Clause permits the prosecution to introduce a forensic laboratory report containing a testimonial certification — made for the purpose of proving a particular fact — through the in-court testimony of a scientist who did not sign the certification or perform or observe the test reported in the certification. We hold that surrogate testimony of that order does not meet the constitutional requirement. The accused’s right is to be confronted with the analyst who made the certification, unless that analyst is unavailable at trial, and the accused had an opportunity, pretrial, to cross-examine that particular scientist.
I
A
In August 2005, a vehicle driven by petitioner Donald Bullcoming rear ended a pickup truck at an intersection in Farmington, New Mexico. When the truckdriver exited his vehicle and approached Bullcoming to exchange insurance information, he noticed that Bullcoming’s eyes were bloodshot. Smelling alcohol on Bullcoming’s breath, the truck-driver told his wife to call the police. Bullcoming left the scene before the police arrived, but was soon apprehended by an officer who observed his performance of field sobriety tests. Upon failing the tests, Bullcoming was arrested for driving a vehicle while “under the influence of intoxicating liquor” (DWI), in violation of N. M. Stat. Ann. § 66-8-102 (2004).
Because Bullcoming refused to take a breath test, the police obtained a warrant authorizing a blood-alcohol analysis. Pursuant to the warrant, a sample of Bullcoming’s blood was drawn at a local hospital. To determine Bullcoming’s blood-alcohol concentration (BAC), the police sent the sample to the New Mexico Department of Health, Scientific Labora*653tory Division (SLD). In a standard SLD form titled “Report of Blood Alcohol Analysis,” participants in the testing were identified, and the forensic analyst certified his finding. App. 62.
' SLD’s report contained in the top block “information . . . filled in by [the] arresting officer.” Ibid, (capitalization omitted). This information included the “reason [the] suspect [was] stopped” (the officer checked “Accident”), and the date (“8.14.05”) and time (“18:25 PM”) the blood sample was drawn. Ibid, (capitalization omitted). The arresting officer also affirmed that he had arrested Bullcoming and witnessed the blood draw. Ibid. The next two blocks contained certifications by the nurse who drew Bullcoming’s blood and the SLD intake employee who received the blood sample sent to the laboratory. Ibid.
Following these segments, the report presented the “certificate of analyst,” ibid, (capitalization omitted), completed and signed by Curtis Caylor, the SLD forensic analyst assigned to test Bullcoming’s blood sample, id., at 62, 64-65. Caylor recorded that the BAC in Bullcoming’s sample was 0.21 grams per hundred milliliters, an inordinately high level. Id., at 62. Caylor also affirmed that “[t]he seal of th[e] sample was received intact and broken in the laboratory,” that “the statements in [the analyst’s block of the report] are correct,” and that he had “followed the procedures set out on the reverse of th[e] report.” Ibid. Those “procedures” instructed analysts, inter alia, to “retai[n] the sample container and the raw data from the analysis,” and to “not[e] any circumstance or condition which might affect the integrity of the sample or otherwise affect the validity of the analysis.” Id., at 65. Finally, in a block headed “certificate of reviewer,” the SLD examiner who reviewed Caylor’s analysis certified that Caylor was qualified to conduct the BAC test, and that the “established procedure” for handling and analyzing Bullcoming’s sample “ha[d] been followed.” Id., at 62 (capitalization omitted).
*654SLD analysts use gas chromatograph machines to determine BAC levels. Operation of the machines requires specialized knowledge and training. Several steps are involved in the gas chromatograph process, and human error can occur at each step.1
*655Caylor’s report that Bullcoming’s BAC was 0.21 supported a prosecution for aggravated DWI, the threshold for which is a BAC of 0.16 grams per hundred milliliters, § 66-8-102(D)(1). The State accordingly charged Bullcoming with this more serious crime.
B
The case was tried to a jury in November 2005, after our decision in Crawford v. Washington, 541 U. S. 36 (2004), but before Melendez-Diaz. On the day of trial, the State announced that it would not be calling SLD analyst Curtis Caylor as a witness because he had “very recently [been] put on unpaid leave” for a reason not revealed. 2010-NMSC-007, ¶ 8, 147 N. M. 487, 492, 226 P. 3d 1, 6 (internal quotation marks omitted); App. 58. A startled defense counsel objected. The prosecution, she complained, had never disclosed, until trial commenced, that the witness “out there ... [was] not the analyst [of Bullcoming’s sample].” Id., at 46. Counsel stated that, “had [she] known that the analyst [who tested Bullcoming’s blood] was not available," her opening, indeed, her entire defense “may very well have been dramatically different.” Id., at 47. The State, however, proposed to introduce Caylor’s finding as a “business record” during the testimony of Gerasimos Razatos, an SLD scientist who had neither observed nor reviewed Caylor’s analysis. Id., at 44.
Bullcoming’s counsel opposed the State’s proposal. Id., at 44-45. Without Caylor’s testimony, defense counsel maintained, introduction of the analyst’s finding would violate Bullcoming’s Sixth Amendment right “to be confronted with *656the witnesses against him.” Ibid.2 The trial court overruled the objection, id., at 46-47, and admitted the SLD report as a business record, id., at 44-46, 57.3 The jury convicted Bullcoming of aggravated DWI, and the New Mexico Court of Appeals upheld the conviction, concluding that “the blood alcohol report in the present case was non-testimonial and prepared routinely with guarantees of trustworthiness.” 2008-NMCA-097, ¶ 17, 144 N. M. 546, 552,189 P. 3d 679, 685.
C
While Bullcoming’s appeal was pending before the New Mexico Supreme Court, this Court decided Melendez-Diaz. In that case, “[t]he Massachusetts courts [had] admitted into evidence affidavits reporting the results of forensic analysis which showed that material seized by the police and connected to the defendant was cocaine.” 557 U. S., at 307. Those affidavits, the Court held, were “ ‘testimonial,’ rendering the affiants ‘witnesses’ subject to the defendant’s right of confrontation under the Sixth Amendment.” Ibid.
In light of Melendez-Diaz, the New Mexico Supreme Court acknowledged that the blood-alcohol report introduced at Bullcoming’s trial qualified as testimonial evidence. Like the affidavits in Melendez-Diaz, the court observed, the report was “functionally identical to live, in-court testimony, doing precisely what a witness does on direct examination.” 147 N. M., at 494, 226 P. 3d, at 8 (quoting Melendez-Diaz, 557 *657U. S., at 310-311).4 Nevertheless, for two reasons, the court held that admission of the report did not violate the Confrontation Clause.
First, the court said certifying analyst Caylor “was a mere scrivener,” who “simply transcribed the results generated by the gas chromatograph machine.” 147 N. M., at 494-495, 226 P. 3d, at 8-9. Second, SLD analyst Razatos, although he did not participate in testing Bullcoming's blood, “qualified as an expert witness with respect to the gas chromato-graph machine.” Id., at 495, 226 P. 3d, at 9. “Razatos provided live, in-court testimony,” the court stated, “and, thus, was available for cross-examination regarding the operation of the . . . machine, the results of [Bullcoming’s] BAC test, and the SLD’s established laboratory procedures.” Ibid. Razatos’ testimony was crucial, the court explained, because Bullcoming could not cross-examine the machine or the written report. Id., at 496,226 P. 3d, at 10. But “[Bullcoming’s] right of confrontation was preserved,” the court concluded, because Razatos was a qualified analyst, able to serve as a surrogate for Caylor. Ibid.
We granted certiorari to address this question: Does the Confrontation Clause permit the prosecution to introduce a forensic laboratory report containing a testimonial certification, made in order to prove a fact at a criminal trial, through the in-court testimony of an analyst who did not sign the certification or personally perform or observe the performance of the test reported in the certification. 561 U. S. 1058 (2010). Our answer is in line with controlling precedent: As a rule, if an out-of-court statement is testimonial in nature, it may not be introduced against the accused at trial unless the witness who made the statement is unavailable and the accused has had a prior opportunity to confront that witness. *658Because the New Mexico Supreme Court permitted the testimonial statement of one witness, i. e., Caylor, to enter into evidence through the in-court testimony of a second person, i. e., Razatos, we reverse that court’s judgment.
I
The Sixth Amendment’s Confrontation Clause confers upon the accused, “[i]n all criminal prosecutions, . . . the right ... to be confronted with the witnesses against him.” In a pathmarking 2004 decision, Crawford v. Washington, we overruled Ohio v. Roberts, 448 U. S. 56 (1980), which had interpreted the Confrontation Clause to allow admission of absent witnesses’ testimonial statements based on a judicial determination of reliability. See id., at 66. Rejecting Roberts’ “amorphous notions of ‘reliability,’ ” Crawford, 541 U. S., at 61, Crawford held that fidelity to the Confrontation Clause permitted admission of “[tjestimonial statements of witnesses absent from trial.. . only where the declarant is unavailable, and only where the defendant has had a prior opportunity to cross-examine,” id., at 59. See Michigan v. Bryant, 562 U. S. 344, 354 (2011) (“[F]or testimonial evidence to be admissible, the Sixth Amendment ‘demands what the common law required: unavailability [of the witness] and a prior opportunity for cross-examination.’” (quoting Crawford, 541 U. S., at 68)). Melendez-Diaz, relying on Crawford’s, rationale, refused to create a “forensic evidence” exception to this rule. 557 U. S., at 317-321.5 An analyst’s certification prepared in connection with a criminal investigation or prosecution, the Court held, is “testimonial,” and *659therefore within the compass of the Confrontation Clause. Id., at 321-324.6
The State in the instant case never asserted that the analyst who signed the certification, Curtis Caylor, was unavailable. The record showed only, that Caylor was placed on unpaid leave for an undisclosed reason. See supra, at 655. Nor did Bullcoming have an opportunity to cross-examine Caylor. Crawford and Melendez-Diaz, therefore, weigh heavily in Bullcoming’s favor. The New Mexico Supreme Court, however, although recognizing that the SLD report was testimonial for purposes of the Confrontation Clause, considered SLD analyst Razatos an adequate substitute for Caylor. We explain first why Razatos’ appearance did not meet the Confrontation Clause requirement. We next address the State’s argument that the SLD report ranks as “nontestimonial,” and therefore “[was] not subject to the Confrontation Clause” in the first place. Brief for Respondent 7 (capitalization omitted).
A
The New Mexico Supreme Court held surrogate testimony adequate to satisfy the Confrontation Clause in this case because analyst Caylor “simply transcribed the resul[t] generated by the gas chromatograph machine,” presenting no interpretation and exercising no independent judgment. 226 R 3d, at 8. Bullcoming’s “true ‘accuser,’” the court said, was the machine, while testing analyst Caylor’s role was that of “mere scrivener.” Id., at 9. Caylor’s certification, how*660ever, reported more than a machine-generated number. See supra, at 653.
Caylor certified that he received Bullcoming's blood sample intact with the seal unbroken, that he checked to make sure that the forensic report number and the sample number “correspond[ed],” and that he performed on Bullcoming’s sample a particular test, adhering to a precise protocol. App. 62-65. He further represented, by leaving the “[Remarks” section of the report blank, that no “circumstance or condition . . . a£fect[ed] the integrity of the sample or . . . the validity of the analysis.” Id., at 62, 65. These representations, relating to past events and human actions not revealed in raw, machine-produced data, are meet for cross-examination.
The potential ramifications of the New Mexico' Supreme Court's reasoning, furthermore, raise red flags. Most witnesses, after all, testify to their observations of factual conditions or events, e. g., “the light was green,” “the hour was noon.” Such witnesses may record, on the spot, what they observed. Suppose a police report recorded an objective fact — Bullcoming’s counsel posited the address above the front door of a house or the readout of a radar gun. See Brief for Petitioner 35. Could an officer other than the one who saw the number on the house or gun present the information in court — so long as that officer was equipped to testify about any technology the observing officer deployed and the police department’s standard operating procedures? As our precedent makes plain, the answer is emphatically “No.” See Davis v. Washington, 547 U. S. 813, 826 (2006) (Confrontation Clause may not be “evaded by having a note-taking police [officer] recite the .. . testimony of the declarant” (emphasis deleted)); Melendez-Diaz, 557 U. S., at 334 (Kennedy, J., dissenting) (“The Court made clear in Davis that it will not permit the testimonial statement of one witness to enter into evidence through the in-court testimony of a second.”).
*661The New Mexico Supreme Court stated that the number registered by the gas chromatograph machine called for no interpretation or exercise of independent judgment on Caylor’s part. 147 N. M., at 494-495, 226 P. 3d, at 8-9. We have already explained that Caylor certified to more than a machine-generated number. See supra, at 653. In any event, the comparative reliability of an analyst’s testimonial report drawn from machine-produced data does not overcome the Sixth Amendment bar. This Court settled in Crawford that the “obviou[s] reliability]” of a testimonial statement does not dispense wiLh the Confrontation Clause. 541 U. S., at 62; see id., at 61 (Clause “commands, not that evidence be reliable, but that reliability be assessed in a particular manner: by testing [the evidence] in the crucible of cross-examination”). Accordingly, the analysts who write reports that the prosecution introduces must be made available for confrontation even if they possess “the scientific acumen of Mme. Curie and the veracity of Mother Teresa.” Melendez-Diaz, 557 U. S., at 319-320, n. 6.
B
Recognizing that admission of the blood-alcohol analysis depended on “live, in-court testimony [by] a qualified analyst,” 147 N. M., at 496, 226 P. 3d, at 10, the New Mexico Supreme Court believed that Razatos could substitute for Caylor because Razatos “qualified as an expert witness with respect to the gas chromatograph machine and the SLD’s laboratory procedures,” id., at 495, 226 P. 3d, at 9. But surrogate testimony of the kind Razatos was equipped to give could not convey what Caylor knew or observed about the events his certification concerned, i. e., the particular test and testing process he employed.7 Nor could such surrogate *662testimony expose any lapses or lies on the certifying analyst's part.8 Significant here, Razatos had no knowledge of the reason why Caylor had been placed on unpaid leave. With Caylor on the stand, Bullcoming’s counsel could have asked questions designed to reveal whether incompetence, evasiveness, or dishonesty accounted for Caylor’s removal from his workstation. Notable in this regard, the State never asserted that Caylor was "unavailable”; the prosecution conveyed only that Caylor was on uncompensated leave. Nor did the State assert that Razatos had any "independent opinion” concerning Bullcoming’s BAC. See Brief for Respondent 58, n. 15. In this light, Caylor’s live testimony could hardly be typed "a hollow formality,” post, at 677.
More fundamentally, as this Court stressed in Crawford, “[t]he text of the Sixth Amendment does not suggest any open-ended exceptions from the confrontation requirement to be developed by the courts.” 541 U. S., at 54. Nor is it “the role of courts to extrapolate from the words of the [Confrontation Clause] to the values behind it, and then to enforce its guarantees only to the extent they serve (in the courts’ views) those underlying values.” Giles v. California, 554 U. S. 353, 375 (2008) (plurality). Accordingly, the Clause does not tolerate dispensing with confrontation simply because the court believes that questioning one witness about another’s testimonial statements provides a fair enough opportunity for cross-examination.
A recent decision involving another Sixth Amendment right — the right to counsel — is instructive. In United States v. Gonzalez-Lopez, 548 U. S. 140 (2006), the Govern*663ment argued that illegitimately denying a defendant his counsel of choice did not violate the Sixth Amendment where “substitute counsel’s performance” did not demonstrably prejudice the defendant. Id., at 144-145. This Court rejected the Government’s argument. “[T]rue enough,” the Court explained, “the purpose of the rights set forth in [the Sixth] Amendment is to ensure a fair trial; but it does not follow that the rights can be disregarded so long as the trial is, on the whole, fair.” Id., at 145. If a “particular guarantee” of the Sixth Amendment is violated, no substitute procedure can cure the violation, and “[n]o additional showing of prejudice is required to make the violation ‘complete.’ ” Id., at 146. If representation by substitute counsel does not satisfy the Sixth Amendment, neither does the opportunity to confront a substitute witness.
In short, when the Slate elected to introduce Gaylor’s certification, Caylor became a witness Bullcoming had the right to confront. Our precedent cannot sensibly be read any other way. See Melendez-Diaz, 557 U. S., at 334 (Kennedy, J., dissenting) (Court’s holding means “the . . . analyst who must testify is the person who signed the certificate”).
III
We turn, finally, to the State s contention that SLD s blood-alcohol analysis reports are nontestimonial in character, therefore no Confrontation Clause question even arises in this case. Melendez-Diaz left no room for that argument, the New Mexico Supreme Court concluded, see 147 N. M., at 494, 226 P. 3d, at 7-8; supra, at 656-657, a conclusion we find inescapable.
In Melendez-Diaz, a state forensic laboratory, on police request, analyzed seized evidence (plastic bags) and reported the laboratory’s analysis to the police (the substance found in the bags contained cocaine). 557 U. S., at 308. The “certificates of analysis” prepared by the analysts who tested the evidence in Melendez-Diaz, this Court held, were *664“incontrovertibly . . . affirmation[s] made for the purpose of establishing or proving some fact” in a criminal proceeding. Id., at 310 (internal quotation marks omitted). The same purpose was served by the certificate in question here.
The State maintains that the affirmations made by analyst Caylor were not “adversarial” or “inquisitorial,” Brief for Respondent 27-33; instead, they were simply observations of an “independent scientis[t]” made “according to a non-adversarial public duty,” id., at 32-33. That argument fares no better here than it did in Melendez-Diaz. A document created solely for an “evidentiary purpose,” Melendez-Diaz clarified, made in aid of a police investigation, ranks as testimonial. 557 U. S., at 311 (forensic reports available for use at trial are “testimonial statements” and certifying analyst is a “‘witnes[s]’ for purposes of the Sixth Amendment”).
Distinguishing Bullcoming’s case from Melendez-Diaz, where the analysts’ findings were contained in certificates “sworn to before a notary public,” id., at 308, the State emphasizes that the SLD report of Bullcoming’s BAC was “un-sworn.” Brief for Respondent 13; post, at 676 (“only sworn statement” here was that of Razatos, “who was present and [did] testif[y]”). As the New Mexico Supreme Court recognized, “ The absence of [an] oath [i]s not dispositive’ in determining if a statement is testimonial.” 147 N. M., at 494, 226 P. 3d, at 8 (quoting Crawford, 541 U. S., at 52). Indeed, in Crawford, this Court rejected as untenable any construction of the Confrontation Clause that would render inadmissible only sworn ex parte affidavits, while leaving admission of formal, but unsworn statements “perfectly OK.” Id., at 52-53, n. 3. Reading the Clause in this “implausible” manner, ibid., the Court noted, would make the right to confrontation easily erasable. See Davis, 547 U. S., at 830-831, n. 5; id., at 838 (Thomas, J., concurring in judgment in part and dissenting in part).
In all material respects, the laboratory report in this case resembles those in Melendez-Diaz. Here, as in Melendez-*665Diaz, a law-enforcement officer provided seized evidence to a state laboratory required by law to assist in police investigations, N. M. Stat. Ann. § 29-3-4 (2004). Like the analysts in Melendez-Diaz, analyst Caylor tested the evidence and prepared a certificate concerning the result of his analysis. App. 62. Like the Melendez-Diaz certificates, Caylor’s certificate is “formalized” in a signed document, Davis, 547 U. S., at 837, n. 2 (opinion of Thomas, J.), headed a “report,” App. 62. Noteworthy as well, the SLD report form contains a legend referring to municipal and magistrate courts’ rules that provide for the admission of certified blood-alcohol analyses.
In sum, the formalities attending the “report of blood alcohol analysis” are more than adequate to qualify Caylor’s assertions as testimonial. The absence of notarization does not remove his certification from Confrontation Clause governance. The New Mexico Supreme Court, guided by Melendez-Diaz, correctly recognized that Caylor’s report ‘Tell within the core class of testimonial statements,” 147 N. M., at 493, 226 P. 3d, at 7, described in this Court’s leading Confrontation Clause decisions: Melendez-Diaz, 557 U. S., at 310; Davis, 547 U. S., at 830; Crawford, 541 U. S., at 51-52.
IV
The State and its amici urge that unbending application of the Confrontation Clause to forensic evidence would impose an undue burden on the prosecution. This argument, also advanced in the dissent, post, at 683, largely repeats a refrain rehearsed and rejected in Melendez-Diaz. See 557 U. S., at 325-328. The constitutional requirement, we reiterate, “may not [be] disregarded] ... at our convenience,” id., at 325, and the predictions of dire consequences, we again observe, are dubious, see ibid.
New Mexico law, it bears emphasis, requires the laboratory to preserve samples, which can be retested by other analysts, see N. M. Admin. Code § 7.33.2.15(A)(4)-(6) (2010), *666available at http://www.nmcpr.state.nm.us/nmac/_title07/ T07C033.htm, and neither party questions SLD’s compliance with that requirement. Retesting “is almost always an option ... in [DWI] cases,” Brief for Public Defender Service for District of Columbia et al. as Amici Curiae 25 (hereinafter PDS Brief), and the State had that option here: New Mexico could have avoided any Confrontation Clause problem by asking Razatos to retest the sample, and then testify to the results of his retest rather than to the results of a test he did not conduct or observe.
Notably, New Mexico advocates retesting as an effective means to preserve a defendant’s confrontation right “when the [out-of-court] statement is raw data or a mere transcription of raw data onto a public record.” Brief for Respondent 53-54. But the State would require the defendant to initiate retesting. Id., at 55; post, at 677 (defense “remains free to . . . call and examine the technician who performed a test”); post, at 681 (“free retesting” is available to defendants). The prosecution, however, bears the burden of proof. Melendez-Diaz, 557 U. S., at 324 (“[T]he Confrontation Clause imposes a burden on the prosecution to present its witnesses, not on the defendant to bring those adverse witnesses into court.”). Hence the obligation to propel retesting when the original analyst is unavailable is the State’s, not the defendant’s. See Taylor v. Illinois, 484 U. S. 400, 410, n. 14 (1988) (Confrontation Clause’s requirements apply “in every case, whether or not the defendant seeks to rebut the case against him or to present a case of his own”).
Furthermore, notice-and-demand procedures, long in effect in many jurisdictions, can reduce burdens on forensic laboratories. Statutes governing these procedures typically “render . . . otherwise hearsay forensic reports admissible[,] while specifically preserving a defendant’s right to demand that the prosecution call the author/analyst of [the] report.” PDS Brief 9; see Melendez-Diaz, 557 U. S., at 326 (observing that notice-and-demand statutes “permit the defendant to assert (or forfeit by silence) his Confrontation Clause right *667after receiving notice of the prosecution’s intent to use a forensic analyst’s report”).
Even before this Court’s decision in Crawford, moreover, it was common prosecutorial practice to call the forensic analyst to testify. Prosecutors did so “to bolster the persuasive power of [the State’s] case[,] . . . [even] when the defense would have preferred that the analyst did not testify.” PDS Brief 8.
We note also the “small fraction of... cases” that “actually proceed to trial.” Melendez-Diaz, 557 U. S., at 325 (citing estimate that “nearly 95% of convictions in state and federal courts are obtained via guilty plea”). And, “when cases in which forensic analysis has been conducted [do] go to trial,” defendants “regularly ... [stipulate] to the admission of [the] analysis.” PDS Brief 20. “[A]s a result, analysts testify in only a very small percentage of cases,” id., at 21, for “[i]t is unlikely that defense counsel will insist on live testimony whose effect will be merely to highlight rather than cast doubt upon the forensic analysis.” Melendez-Diaz, 557 U. S., at 328.9
Tellingly, in jurisdictions in which “it is the [acknowledged] job of . . . analysts to testify in court. . . about their test results,” the sky has not fallen. PDS Brief 23. State *668and municipal laboratories “make operational and staffing decisions” to facilitate analysts’ appearance at trial. Ibid. Prosecutors schedule trial dates to accommodate analysts’ availability, and trial courts liberally grant continuances when unexpected conflicts arise. Id., at 24-25. In rare cases in which the analyst is no longer employed by the laboratory at the time of trial, “the prosecution makes the effort to bring that analyst ... to court.” Id., at 25. And, as is the practice in New Mexico, see supra, at 665-666, laboratories ordinarily retain additional samples, enabling them to run tests again when necessary.10
H» ^
For the reasons stated, the judgment of the New Mexico Supreme Court is reversed, and the case is remanded for further proceedings not inconsistent with this opinion.11
It is so ordered.
Justice Sotomayor and Justice Kagan join all but Part IV of this opinion. Justice Thomas joins all but Part IV and footnote 6.
Gas chromatography is a widely used scientific method of quantitatively analyzing the constituents of a mixture. See generally H. McNair & J. Miller, Basic Gas Chromatography (2d ed. 2009) (hereinafter McNair). Under SLD’s standard testing protocol, the analyst extracts two blood samploo and inserto them into vialo containing an “internal standard” — a chemical additive. App. 53. See McNair 141-142. The analyst then “cap[s] the [two] sample[s],” “erimp[s] them with an aluminum top,” and placeo the vialo into the gaa chromatograph machine. App. 53-54. Within a few hours, thio device produces a printed graph a chromatogram — along with calculations representing a software generated inter pretation of the data. See Brief for New Mexico Department of Health, SLD, as Amicus Curiae 16-17.
Although the State presented testimony that obtaining an accurate BAC measurement merely entails “look[ing] at the [gas chromatograph] machine and rocord[ing] the resultn,” App. 54, authoritative sources reveal that the matter is not so simple or certin. “In order to perform quantita tive analyses satisfactorily and . . . support the results under rigorous examination in court, the analyst must be aware of, and adhere to, good analytical prácticos and understand what is being done and why.” Stafford, Chromatography, in Principles of Forensic Toxicology 91, 114 (B. Levine 2d ed. 2006). See also McNair 137 (“Errors that occur in any step can invalidate the best chromatographic analysis, go attention must bo paid to all steps.”); D. Bartell, M. McMurray, & A. ImObersteg, Attacking and Defending Drunk Driving Tests §16:80 (2d revision 2010) (stating that 93% of erroro in laboratory tests for BAC levels arc human errora that occur cither before or after machines analyze samples). Even after the machine has produced its printed result, a review of the chromatogram may indicate that the test was not valid. See McNair 207-214.
Nor is the risk of human error so remote as to be negligible. Amici inform usj for example, that in neighboring Colorado, a single forensic laboratory produced at least 206 flawed blood alcohol readings over a throe-year span, prompting the diamisoal of several criminal prosecutions. See Brief for National Association of Criminal Defense Lawyers et al. as Amici Curiae 32-33. An analyst had used improper amounts of the internal standard, causing the chromatograph machine systematically to in-*655fiate BAC measurements. The analyst’s error, a supervisor said, was “fairly complex.” Ensslin, Final Tally on Flawed DUI: 206 Errors, 9 Tossed or Reduced, Colorado Springs Gazette, Apr. 19, 2010, pp. 1,2 (internal quotation marko omitted), available at http://www.gazette.com/articloo/ report-97354-police-discuss.html. (All Internet materials as visited June 21, 2011, and included in Clerk of Court’s ease file.)
The State called as witnesses the arresting officer and the nurse who drew Bullcoming’s blood. Bullcoming did not object to the State’s failure to call the SLD intake employee or the reviewing analyst. “It is up to the prosecution,” the Court observed in Melendez-Diaz v. Massachusetts, 557 U. S. 305, 311, n. 1 (2009), “to decide what steps in the chain of custody arc so crucial as to require evidence; but what testimony ia introduced must (if the defendant objects) be introduced live.”
The trial judge noted that, when he started out in law practice, “there were no breath tests or blood tests. They just brought in the cop, and the cop said, ‘Yeah, he was drunk.’” App. 47.
The dissent makes plain that its objection is less to the application of the Court’s docisiono in Crawford and Melendez-Diaz to this case than to those pathmarking decisions themselves. See post, at 678 (criticizing the Crawford “line of cases” for rejecting “reliable evidence”); post, at 681, 684 (deploring “Crawford’s rejection of the [reliability-centered] regime of Ohio v. Roberts”).
In so ruling, the New Mexico Supreme Court explicitly overruled State v. Dedman, 2004-NMSC-037, 136 N. M. 561, 102 P. 3d 628 (2004), which had classified blood-alcohol reports as public records neither “investigative nor prosecutorial” in nature. 147 N. M., at 494, 226 P. 3d, at 7-8.
To rank as “testimonial,” a statement must have a “primary purpose” of “establish[ing] or prov[ing] past events potentially relevant to later criminal prosecution.” Davis v. Washington, 547 U. S. 813, 822 (2006). See also Bryant, 562 U. S., at 358. Elaborating on the purpose for which a “testimonial report” is created, we observed in Melendez-Dias that business and public records “are generally admissible absent confrontation ... because—having been created for the administration of an entity's affaire and not for the purpose of establishing or proving some fact at trial — they are not testimonial.” 557 U. S., at 324.
We do not question that analyot Caylor, in common with other analysts employed by SLD, likely would not recall a particular test, given the number of tests each analyst conducts and the standard procedure followed in *662tosting, Even do, Caylor’o testimony under oath would havo enabled Bull coming’s counsel to mice before a jury questions concerning Caylor’o pro fieiency, the care he took in performing his work, and his veracity. In particular, Bulleoming’s counsel likely would have inquired on erooD examination why Caylor had been placed on unpaid leave.
At Bullcoming’s trial, Rosatos acknowledged that “you don’t know un-Iodo you actually observe the analysis that someone else conducts, whether they followed th[e] protocol in every instance.” App. 59.
The dissent argues otherwise, reporting a 71% increase, from 2008 to 2010, in the number of subpoenas for New Mexico analysts’ testimony in impaired-driving cases. Post, at 683. The dissent is silent, however, on the number of instances in which subpoenaed analysts in fact testify, i. e., the figure that would reveal the actual burden of courtroom testimony. Moreover, New Mexico's Department of Health, Scientific Laboratory Division, has attributed the “chaotic” conditions noted by the dissent, post, at 684, to several factors, among them, staff attrition, a state hiring freeze, a 15% increase in the number of blood samples received for testing, and “wildly” divergent responses by New Mexico District Attorneys to Melendez-Diaz. Brief for New Mexico Department of Health, SLD, as Amicus Curiae 2-5. Some New Mexico District Attorneys’ offices, we are informed, “subpoen[a] every analyst with any connection to a blood sample,” id., at 5, an exorbitant practice that undoubtedly inflates the number of subpoenas issued.
As in Melendez-Diaz, 557 U. S., at 329, and n. 14, we express no view on whether the Confrontation Clause error in this case wais harmless. The New Mexico Supreme Court did not reach that question, see Brief for Respondent 59-60, and nothing in this opinion impedes a harmless-error inquiry on remand-
The dissent refers, selectively, to experience in Los Angeles, post, at 683, but overlooks experience documented in Michigan. In that State, post-Melendez-Diaz, the increase in in-court analyst testimony has been slight. Compare PDS Brief 21 (in 2006, analysts provided testimony for only 0.7% of all tests) with Michigan State Police, Forensic Science Division, available at http://www.miehigan.gov/msp/ 0,1607,7-123-1593_3800-15901 — ,00.html (in 2010, analysts provided testimony for approximately 1% of all tests).