Arizona v. Youngblood

488 U.S. 51 (1988) ARIZONA v. YOUNGBLOOD No. 86-1904. Supreme Court of United States. Argued October 11, 1988 Decided November 29, 1988 CERTIORARI TO THE COURT OF APPEALS OF ARIZONA *52 John R. Gustafson argued the cause for petitioner. With him on the brief were Stephen D. Neely, James M. Howard, and Deborah Strange Ward. Daniel F. Davis argued the cause and filed a brief for respondent. CHIEF JUSTICE REHNQUIST delivered the opinion of the Court. Respondent Larry Youngblood was convicted by a Pima County, Arizona, jury of child molestation, sexual assault, and kidnaping. The Arizona Court of Appeals reversed his conviction on the ground that the State had failed to preserve semen samples from the victim's body and clothing. 153 Ariz. 50, 734 P. 2d 592 (1986). We granted certiorari to consider the extent to which the Due Process Clause of the Fourteenth Amendment requires the State to preserve evidentiary material that might be useful to a criminal defendant. On October 29, 1983, David L., a 10-year-old boy, attended a church service with his mother. After he left the service at about 9:30 p.m., the boy went to a carnival behind the church, where he was abducted by a middle-aged man of medium height and weight. The assailant drove the boy to a secluded area near a ravine and molested him. He then took the boy to an unidentified, sparsely furnished house where he sodomized the boy four times. Afterwards, the assailant tied the boy up while he went outside to start his car. Once the assailant started the car, albeit with some difficulty, he returned to the house and again sodomized the boy. The assailant then sent the boy to the bathroom to wash up before he returned him to the carnival. He threatened to kill the boy if he told anyone about the attack. The entire ordeal lasted about 1 1/2 hours. After the boy made his way home, his mother took him to Kino Hospital. At the hospital, a physician treated the boy for rectal injuries. The physician also used a "sexual assault kit" to collect evidence of the attack. The Tucson Police Department *53 provided such kits to all hospitals in Pima County for use in sexual assault cases. Under standard procedure, the victim of a sexual assault was taken to a hospital, where a physician used the kit to collect evidence. The kit included paper to collect saliva samples, a tube for obtaining a blood sample, microscopic slides for making smears, a set of Q-Tip-like swabs, and a medical examination report. Here, the physician used the swab to collect samples from the boy's rectum and mouth. He then made a microscopic slide of the samples. The doctor also obtained samples of the boy's saliva, blood, and hair. The physician did not examine the samples at any time. The police placed the kit in a secure refrigerator at the police station. At the hospital, the police also collected the boy's underwear and T-shirt. This clothing was not refrigerated or frozen. Nine days after the attack, on November 7, 1983, the police asked the boy to pick out his assailant from a photographic lineup. The boy identified respondent as the assailant. Respondent was not located by the police until four weeks later; he was arrested on December 9, 1983. On November 8, 1983, Edward Heller, a police criminologist, examined the sexual assault kit. He testified that he followed standard department procedure, which was to examine the slides and determine whether sexual contact had occurred. After he determined that such contact had occurred, the criminologist did not perform any other tests, although he placed the assault kit back in the refrigerator. He testified that tests to identify blood group substances were not routinely conducted during the initial examination of an assault kit and in only about half of all cases in any event. He did not test the clothing at this time. Respondent was indicted on charges of child molestation, sexual assault, and kidnaping. The State moved to compel respondent to provide blood and saliva samples for comparison with the material gathered through the use of the sexual assault kit, but the trial court denied the motion on the *54 ground that the State had not obtained a sufficiently large semen sample to make a valid comparison. The prosecutor then asked the State's criminologist to perform an ABO blood group test on the rectal swab sample in an attempt to ascertain the blood type of the boy's assailant. This test failed to detect any blood group substances in the sample. In January 1985, the police criminologist examined the boy's clothing for the first time. He found one semen stain on the boy's underwear and another on the rear of his T-shirt. The criminologist tried to obtain blood group substances from both stains using the ABO technique, but was unsuccessful. He also performed a P-30 protein molecule test on the stains, which indicated that only a small quantity of semen was present on the clothing; it was inconclusive as to the assailant's identity. The Tucson Police Department had just begun using this test, which was then used in slightly more than half of the crime laboratories in the country. Respondent's principal defense at trial was that the boy had erred in identifying him as the perpetrator of the crime. In this connection, both a criminologist for the State and an expert witness for respondent testified as to what might have been shown by tests performed on the samples shortly after they were gathered, or by later tests performed on the samples from the boy's clothing had the clothing been properly refrigerated. The court instructed the jury that if they found the State had destroyed or lost evidence, they might "infer that the true fact is against the State's interest." 10 Tr. 90. The jury found respondent guilty as charged, but the Arizona Court of Appeals reversed the judgment of conviction. It stated that " `when identity is an issue at trial and the police permit the destruction of evidence that could eliminate the defendant as the perpetrator, such loss is material to the defense and is a denial of due process.' " 153 Ariz., at 54, 734 P. 2d, at 596, quoting State v. Escalante, 153 Ariz. 55, 61, 734 P. 2d 597, 603 (App. 1986). The Court of Appeals *55 concluded on the basis of the expert testimony at trial that timely performance of tests with properly preserved semen samples could have produced results that might have completely exonerated respondent. The Court of Appeals reached this conclusion even though it did "not imply any bad faith on the part of the State." 153 Ariz., at 54, 734 P. 2d, at 596. The Supreme Court of Arizona denied the State's petition for review, and we granted certiorari. 485 U. S. 903 (1988). We now reverse. Decision of this case requires us to again consider "what might loosely be called the area of constitutionally guaranteed access to evidence." United States v. Valenzuela-Bernal, 458 U. S. 858, 867 (1982). In Brady v. Maryland, 373 U. S. 83 (1963), we held that "the suppression by the prosecution of evidence favorable to the accused upon request violates due process where the evidence is material either to guilt or to punishment, irrespective of the good faith or bad faith of the prosecution." Id., at 87. In United States v. Agurs, 427 U. S. 97 (1976), we held that the prosecution had a duty to disclose some evidence of this description even though no requests were made for it, but at the same time we rejected the notion that a "prosecutor has a constitutional duty routinely to deliver his entire file to defense counsel." Id., at 111; see also Moore v. Illinois, 408 U. S. 786, 795 (1972) ("We know of no constitutional requirement that the prosecution make a complete and detailed accounting to the defense of all police investigatory work on a case"). There is no question but that the State complied with Brady and Agurs here. The State disclosed relevant police reports to respondent, which contained information about the existence of the swab and the clothing, and the boy's examination at the hospital. The State provided respondent's expert with the laboratory reports and notes prepared by the police criminologist, and respondent's expert had access to the swab and to the clothing. *56 If respondent is to prevail on federal constitutional grounds, then, it must be because of some constitutional duty over and above that imposed by cases such as Brady and Agurs. Our most recent decision in this area of the law, California v. Trombetta, 467 U. S. 479 (1984), arose out of a drunken driving prosecution in which the State had introduced test results indicating the concentration of alcohol in the blood of two motorists. The defendants sought to suppress the test results on the ground that the State had failed to preserve the breath samples used in the test. We rejected this argument for several reasons: first, "the officers here were acting in `good faith and in accord with their normal practice,' " id., at 488, quoting Killian v. United States, 368 U. S. 231, 242 (1961); second, in the light of the procedures actually used the chances that preserved samples would have exculpated the defendants were slim, 467 U. S., at 489; and, third, even if the samples might have shown inaccuracy in the tests, the defendants had "alternative means of demonstrating their innocence." Id., at 490. In the present case, the likelihood that the preserved materials would have enabled the defendant to exonerate himself appears to be greater than it was in Trombetta, but here, unlike in Trombetta, the State did not attempt to make any use of the materials in its own case in chief.[*] *57 Our decisions in related areas have stressed the importance for constitutional purposes of good or bad faith on the part of the Government when the claim is based on loss of evidence attributable to the Government. In United States v. Marion, 404 U. S. 307 (1971), we said that "[n]o actual prejudice to the conduct of the defense is alleged or proved, and there is no showing that the Government intentionally delayed to gain some tactical advantage over appellees or to harass them." Id., at 325; see also United States v. Lovasco, 431 U. S. 783, 790 (1977). Similarly, in United States v. Valenzuela-Bernal, supra, we considered whether the Government's deportation of two witnesses who were illegal aliens violated due process. We held that the prompt deportation of the witnesses was justified "upon the Executive's good-faith determination that they possess no evidence favorable to the defendant in a criminal prosecution." Id., at 872. The Due Process Clause of the Fourteenth Amendment, as interpreted in Brady, makes the good or bad faith of the State irrelevant when the State fails to disclose to the defendant material exculpatory evidence. But we think the Due Process Clause requires a different result when we deal with the failure of the State to preserve evidentiary material of which no more can be said than that it could have been subjected to tests, the results of which might have exonerated the defendant. Part of the reason for the difference in treatment is found in the observation made by the Court in Trombetta, supra, at 486, that "[w]henever potentially exculpatory *58 evidence is permanently lost, courts face the treacherous task of divining the import of materials whose contents are unknown and, very often, disputed." Part of it stems from our unwillingness to read the "fundamental fairness" requirement of the Due Process Clause, see Lisenba v. California, 314 U. S. 219, 236 (1941), as imposing on the police an undifferentiated and absolute duty to retain and to preserve all material that might be of conceivable evidentiary significance in a particular prosecution. We think that requiring a defendant to show bad faith on the part of the police both limits the extent of the police's obligation to preserve evidence to reasonable bounds and confines it to that class of cases where the interests of justice most clearly require it, i. e., those cases in which the police themselves by their conduct indicate that the evidence could form a basis for exonerating the defendant. We therefore hold that unless a criminal defendant can show bad faith on the part of the police, failure to preserve potentially useful evidence does not constitute a denial of due process of law. In this case, the police collected the rectal swab and clothing on the night of the crime; respondent was not taken into custody until six weeks later. The failure of the police to refrigerate the clothing and to perform tests on the semen samples can at worst be described as negligent. None of this information was concealed from respondent at trial, and the evidence — such as it was — was made available to respondent's expert who declined to perform any tests on the samples. The Arizona Court of Appeals noted in its opinion — and we agree — that there was no suggestion of bad faith on the part of the police. It follows, therefore, from what we have said, that there was no violation of the Due Process Clause. The Arizona Court of Appeals also referred somewhat obliquely to the State's "inability to quantitatively test" certain semen samples with the newer P-30 test. 153 Ariz., at 54, 734 P. 2d, at 596. If the court meant by this statement *59 that the Due Process Clause is violated when the police fail to use a particular investigatory tool, we strongly disagree. The situation here is no different than a prosecution for drunken driving that rests on police observation alone; the defendant is free to argue to the finder of fact that a breathalyzer test might have been exculpatory, but the police do not have a constitutional duty to perform any particular tests. The judgment of the Arizona Court of Appeals is reversed, and the case is remanded for further proceedings not inconsistent with this opinion. Reversed. JUSTICE STEVENS, concurring in the judgment. Three factors are of critical importance to my evaluation of this case. First, at the time the police failed to refrigerate the victim's clothing, and thus negligently lost potentially valuable evidence, they had at least as great an interest in preserving the evidence as did the person later accused of the crime. Indeed, at that time it was more likely that the evidence would have been useful to the police — who were still conducting an investigation — and to the prosecutor — who would later bear the burden of establishing guilt beyond a reasonable doubt — than to the defendant. In cases such as this, even without a prophylactic sanction such as dismissal of the indictment, the State has a strong incentive to preserve the evidence. Second, although it is not possible to know whether the lost evidence would have revealed any relevant information, it is unlikely that the defendant was prejudiced by the State's omission. In examining witnesses and in her summation, defense counsel impressed upon the jury the fact that the State failed to preserve the evidence and that the State could have conducted tests that might well have exonerated the defendant. See App. to Pet. for Cert. C21-C38, C42-C45; 9 Tr. 183-202, 207-208; 10 Tr. 58-61, 69-70. More significantly, the trial judge instructed the jury: "If you find that the State has . . . allowed to be destroyed or lost any evidence whose *60 content or quality are in issue, you may infer that the true fact is against the State's interest." 10 Tr. 90. As a result, the uncertainty as to what the evidence might have proved was turned to the defendant's advantage. Third, the fact that no juror chose to draw the permissive inference that proper preservation of the evidence would have demonstrated that the defendant was not the assailant suggests that the lost evidence was "immaterial." Our cases make clear that "[t]he proper standard of materiality must reflect our overriding concern with the justice of the finding of guilt," and that a State's failure to turn over (or preserve) potentially exculpatory evidence therefore "must be evaluated in the context of the entire record." United States v. Agurs, 427 U. S. 97, 112 (1976) (footnotes omitted); see also California v. Trombetta, 467 U. S. 479, 488 (1984) (duty to preserve evidence "must be limited to evidence that might be expected to play a significant role in the suspect's defense"). In declining defense counsel's and the court's invitations to draw the permissive inference, the jurors in effect indicated that, in their view, the other evidence at trial was so overwhelming that it was highly improbable that the lost evidence was exculpatory. In Trombetta, this Court found no due process violation because "the chances [were] extremely low that preserved [breath] samples would have been exculpatory." Id., at 489. In this case, the jury has already performed this calculus based on its understanding of the evidence introduced at trial. Presumably, in a case involving a closer question as to guilt or innocence, the jurors would have been more ready to infer that the lost evidence was exculpatory. With these factors in mind, I concur in the Court's judgment. I do not, however, join the Court's opinion because it announces a proposition of law that is much broader than necessary to decide this case. It states that "unless a criminal defendant can show bad faith on the part of the police, failure to preserve potentially useful evidence does not constitute a *61 denial of due process of law." Ante, at 58. In my opinion, there may well be cases in which the defendant is unable to prove that the State acted in bad faith but in which the loss or destruction of evidence is nonetheless so critical to the defense as to make a criminal trial fundamentally unfair. This, however, is not such a case. Accordingly, I concur in the judgment. JUSTICE BLACKMUN, with whom JUSTICE BRENNAN and JUSTICE MARSHALL join, dissenting. The Constitution requires that criminal defendants be provided with a fair trial, not merely a "good faith" try at a fair trial. Respondent here, by what may have been nothing more than police ineptitude, was denied the opportunity to present a full defense. That ineptitude, however, deprived respondent of his guaranteed right to due process of law. In reversing the judgment of the Arizona Court of Appeals, this Court, in my view, misreads the import of its prior cases and unduly restricts the protections of the Due Process Clause. An understanding of due process demonstrates that the evidence which was allowed to deteriorate was "constitutionally material," and that its absence significantly prejudiced respondent. Accordingly, I dissent. I The Court, with minimal reference to our past cases and with what seems to me to be less than complete analysis, announces that "unless a criminal defendant can show bad faith on the part of police, failure to preserve potentially useful evidence does not constitute a denial of due process of law." Ante, at 58. This conclusion is claimed to be justified because it limits the extent of police responsibility "to that class of cases where the interests of justice most clearly require it, i. e., those cases in which the police themselves by their conduct indicate that the evidence could form a basis for exonerating the defendant." Ibid. The majority has identified clearly one type of violation, for police action affirmatively *62 aimed at cheating the process undoubtedly violates the Constitution. But to suggest that this is the only way in which the Due Process Clause can be violated cannot be correct. Regardless of intent or lack thereof, police action that results in a defendant's receiving an unfair trial constitutes a deprivation of due process. The Court's most recent pronouncement in "what might loosely be called the area of constitutionally guaranteed access to evidence," United States v. Valenzuela-Bernal, 458 U. S. 858, 867 (1982), is in California v. Trombetta, 467 U. S. 479 (1984). Trombetta addressed "the question whether the Amendment . . . demands that the State preserve potentially exculpatory evidence on behalf of defendants." Id., at 481. JUSTICE MARSHALL, writing for the Court, noted that while the particular question was one of first impression, the general standards to be applied had been developed in a number of cases, including Brady v. Maryland, 373 U. S. 83 (1963), and United States v. Agurs, 427 U. S. 97 (1976).[1] Those *63 cases in no way require that government actions that deny a defendant access to material evidence be taken in bad faith in order to violate due process. As noted by the majority, ante, at 55, the Court in Brady ruled 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, irrespective of the good faith or bad faith of the prosecution." 373 U. S., at 87. The Brady Court went on to explain that the principle underlying earlier cases, e. g., Mooney v. Holohan, 294 U. S. 103 (1935) (violation of due process when prosecutor presented perjured testimony), is "not punishment of society for misdeeds of a prosecutor but avoidance of an unfair trial to the accused." 373 U. S., at 87. The failure to turn over material evidence "casts the prosecutor in the role of an architect of a proceeding that does not comport with standards of justice, even though, as in the present case, his action is not `the result of guile.' " Id., at 88 (quoting lower court opinion). In Trombetta, the Court also relied on United States v. Agurs, 427 U. S., at 107, which required a prosecutor to turn over to the defense evidence that was "clearly supportive of a claim of innocence" even without a defense request. The Court noted that the prosecutor's duty was not one of constitutional dimension unless the evidence was such that its "omission deprived the defendant of a fair trial," id., at 108, and explained: "Nor do we believe the constitutional obligation is measured by the moral culpability, or the willfulness, of the prosecutor. If evidence highly probative of innocence is in his file, he should be presumed to recognize its significance even if he has actually overlooked it. . . . If the suppression of evidence results in constitutional error, it is because of the character of the evidence, not *64 the character of the prosecutor." Id., at 110 (footnote omitted).[2] Agurs thus made plain that the prosecutor's state of mind is not determinative. Rather, the proper standard must focus on the materiality of the evidence, and that standard "must reflect our overriding concern with the justice of the finding of guilt." Id., at 112.[3] Brady and Agurs could not be more clear in their holdings that a prosecutor's bad faith in interfering with a defendant's access to material evidence is not an essential part of a due process violation. Nor did Trombetta create such a requirement. Trombetta's initial discussion focused on the due process requirement "that criminal defendants be afforded a meaningful opportunity to present a complete defense," 467 U. S., at 485, and then noted that the delivery of exculpatory evidence to the defendant "protect[s] the innocent from erroneous *65 conviction and ensur[es] the integrity of our criminal justice system." Ibid. Although the language of Trombetta includes a quotation in which the words "in good faith" appear, those words, for two reasons, do not have the significance claimed for them by the majority. First, the words are the antecedent part of the fuller phrase "in good faith and in accord with their normal practice." Id., at 488. That phrase has its source in Killian v. United States, 368 U. S. 231, 242 (1961), where the Court held that the practice of discarding investigators' notes, used to compile reports that were then received in evidence, did not violate due process.[4] In both Killian and Trombetta, the importance of police compliance with usual procedures was manifest. Here, however, the same standard of conduct cannot be claimed. There has been no suggestion that it was the usual procedure to ignore the possible deterioration of important evidence, or generally to treat material evidence in a negligent or reckless manner. Nor can the failure to refrigerate the clothing be squared with the careful steps taken to preserve the sexual-assault kit. The negligent or reckless failure to preserve important evidence just cannot be "in accord with . . . normal practice." Second, and more importantly, Trombetta demonstrates that the absence of bad faith does not end the analysis. The determination in Trombetta that the prosecution acted in good faith and according to normal practice merely prefaced the primary inquiry, which centers on the "constitutional materiality" of the evidence itself. 467 U. S., at 489. There is *66 nothing in Trombetta that intimates that good faith alone should be the measure.[5] The cases in this area clearly establish that police actions taken in bad faith are not the only species of police conduct that can result in a violation of due process. As Agurs points out, it makes no sense to overturn a conviction because a malicious prosecutor withholds information that he mistakenly believes to be material, but which actually would have been of no help to the defense. 427 U. S., at 110. In the same way, it makes no sense to ignore the fact that a defendant has been denied a fair trial because the State allowed evidence that was material to the defense to deteriorate beyond the point of usefulness, simply because the police were inept rather than malicious. I also doubt that the "bad faith" standard creates the bright-line rule sought by the majority. Apart from the inherent difficulty a defendant would have in obtaining evidence to show a lack of good faith, the line between "good faith" and "bad faith" is anything but bright, and the majority's formulation may well create more questions than it answers. What constitutes bad faith for these purposes? Does a defendant have to show actual malice, or would recklessness, or the deliberate failure to establish standards for maintaining and preserving evidence, be sufficient? Does "good faith police work" require a certain minimum of diligence, or will a lazy officer, who does not walk the few extra steps to the evidence refrigerator, be considered to be acting in good faith? While the majority leaves these questions for *67 another day, its quick embrace of a "bad faith" standard has not brightened the line; it only has moved the line so as to provide fewer protections for criminal defendants. II The inquiry the majority eliminates in setting up its "bad faith" rule is whether the evidence in question here was "constitutionally material," so that its destruction violates due process. The majority does not say whether "evidentiary material of which no more can be said than that it could have been subjected to tests, the results of which might have exonerated the defendant," ante, at 57, is, for purposes of due process, material. But because I do not find the question of lack of bad faith dispositive, I now consider whether this evidence was such that its destruction rendered respondent's trial fundamentally unfair. Trombetta requires that a court determine whether the evidence possesses "an exculpatory value that was apparent before the evidence was destroyed," and whether it was "of such a nature that the defendant would be unable to obtain comparable evidence by other reasonably available means." 467 U. S., at 489. In Trombetta neither requirement was met. But it is important to note that the facts of Trombetta differed significantly from those of this case. As such, while the basic standards set by Trombetta are controlling, the inquiry here must be more finely tuned. In Trombetta, samples of breath taken from suspected drunk drivers had been discarded after police had tested them using an Intoxilyzer, a highly accurate and reliable device for measuring blood-alcohol concentration levels. Id., at 481-482. The Court reasoned that the likelihood of the posttest samples proving to be exculpatory was extremely low, and further observed that the defendants were able to attack the reliability of the test results by presenting evidence of the ways in which the Intoxilyzer might have malfunctioned. This case differs from Trombetta in that here no *68 conclusive tests were performed on the relevant evidence. There is a distinct possibility in this case, one not present in Trombetta, that a proper test would have exonerated respondent, unrebutted by any other conclusive test results. As a consequence, although the discarded evidence in Trombetta had impeachment value (i. e., it might have shown that the test results were incorrect), here what was lost to the respondent was the possibility of complete exoneration. Trombetta's specific analysis, therefore, is not directly controlling. The exculpatory value of the clothing in this case cannot be determined with any certainty, precisely because the police allowed the samples to deteriorate. But we do know several important things about the evidence. First, the semen samples on the clothing undoubtedly came from the assailant. Second, the samples could have been tested, using technology available and in use at the local police department, to show either the blood type of the assailant, or that the assailant was a nonsecreter, i. e., someone who does not secrete a blood-type "marker" into other body fluids, such as semen. Third, the evidence was clearly important. A semen sample in a rape case where identity is questioned is always significant. See Hilliard v. Spalding, 719 F. 2d 1443, 1446-1447 (CA9 1983); People v. Nation, 26 Cal. 3d 169, 176-177, 604 P. 2d 1051, 1054-1055 (1980). Fourth, a reasonable police officer should have recognized that the clothing required refrigeration. Fifth, we know that an inconclusive test was done on the swab. The test suggested that the assailant was a nonsecreter, although it was equally likely that the sample on the swab was too small for accurate results to be obtained. And, sixth, we know that respondent is a secreter. If the samples on the clothing had been tested, and the results had shown either the blood type of the assailant or that the assailant was a nonsecreter, its constitutional materiality would be clear. But the State's conduct has deprived the defendant, and the courts, of the opportunity to determine with certainty the import of this evidence: it has "interfere[d] with *69 the accused's ability to present a defense by imposing on him a requirement which the government's own actions have rendered impossible to fulfill." Hilliard v. Spalding, 719 F. 2d, at 1446. Good faith or not, this is intolerable, unless the particular circumstances of the case indicate either that the evidence was not likely to prove exculpatory, or that the defendant was able to use effective alternative means to prove the point the destroyed evidence otherwise could have made. I recognize the difficulties presented by such a situation.[6] The societal interest in seeing criminals punished rightly requires that indictments be dismissed only when the unavailability of the evidence prevents the defendant from receiving a fair trial. In a situation where the substance of the lost evidence is known, the materiality analysis laid out in Trombetta is adequate. But in a situation like the present one, due process requires something more. Rather than allow a State's ineptitude to saddle a defendant with an impossible burden, a court should focus on the type of evidence, the possibility it might prove exculpatory, and the existence of other evidence going to the same point of contention in determining whether the failure to preserve the evidence in question violated due process. To put it succinctly, where no comparable evidence is likely to be available to the defendant, police must preserve physical evidence of a type that they reasonably should know has the potential, if tested, to reveal immutable characteristics of the criminal, and hence to exculpate a defendant charged with the crime. *70 The first inquiry under this standard concerns the particular evidence itself. It must be of a type which is clearly relevant, a requirement satisfied, in a case where identity is at issue, by physical evidence which has come from the assailant. Samples of blood and other body fluids, fingerprints, and hair and tissue samples have been used to implicate guilty defendants, and to exonerate innocent suspects. This is not to say that all physical evidence of this type must be preserved. For example, in a case where a blood sample is found, but the circumstances make it unclear whether the sample came from the assailant, the dictates of due process might not compel preservation (although principles of sound investigation might certainly do so). But in a case where there is no doubt that the sample came from the assailant, the presumption must be that it be preserved. A corollary, particularly applicable to this case, is that the evidence embody some immutable characteristic of the assailant which can be determined by available testing methods. So, for example, a clear fingerprint can be compared to the defendant's fingerprints to yield a conclusive result; a blood sample, or a sample of body fluid which contains blood markers, can either completely exonerate or strongly implicate a defendant. As technology develops, the potential for this type of evidence to provide conclusive results on any number of questions will increase. Current genetic testing measures, frequently used in civil paternity suits, are extraordinarily precise. See Clark v. Jeter, 486 U. S. 456, 465 (1988). The importance of these types of evidence is indisputable, and requiring police to recognize their importance is not unreasonable. The next inquiry is whether the evidence, which was obviously relevant and indicates an immutable characteristic of the actual assailant, is of a type likely to be independently exculpatory. Requiring the defendant to prove that the particular piece of evidence probably would be independently exculpatory *71 would require the defendant to prove the content of something he does not have because of the State's misconduct. Focusing on the type of evidence solves this problem. A court will be able to consider the type of evidence and the available technology, as well as the circumstances of the case, to determine the likelihood that the evidence might have proved to be exculpatory. The evidence must also be without equivalent in the particular case. It must not be cumulative or collateral, cf. United States v. Agurs, 427 U. S., at 113-114, and must bear directly on the question of innocence or guilt. Due process must also take into account the burdens that the preservation of evidence places on the police. Law enforcement officers must be provided the option, as is implicit in Trombetta, of performing the proper tests on physical evidence and then discarding it.[7] Once a suspect has been arrested the police, after a reasonable time, may inform defense counsel of plans to discard the evidence. When the defense has been informed of the existence of the evidence, after a reasonable time the burden of preservation may shift to the defense. There should also be flexibility to deal with evidence that is unusually dangerous or difficult to store. III Applying this standard to the facts of this case, I conclude that the Arizona Court of Appeals was correct in overturning respondent's conviction. The clothing worn by the victim contained samples of his assailant's semen. The appeals court found that these samples would probably be larger, less contaminated, and more likely to yield conclusive test results than would the samples collected by use of the assault kit. 153 Ariz. 50, 54, 734 P. 2d 592, 596 (1986). The clothing *72 and the semen stains on the clothing therefore obviously were material. Because semen is a body fluid which could have been tested by available methods to show an immutable characteristic of the assailant, there was a genuine possibility that the results of such testing might have exonerated respondent. The only evidence implicating respondent was the testimony of the victim.[8] There was no other eyewitness, and the only other significant physical evidence, respondent's car, was seized by police, examined, turned over to a wrecking company, and then dismantled without the victim's having viewed it. The police also failed to check the car to confirm or refute elements of the victim's testimony.[9] *73 Although a closer question, there was no equivalent evidence available to respondent. The swab contained a semen sample, but it was not sufficient to allow proper testing. Respondent had access to other evidence tending to show that he was not the assailant, but there was no other evidence that would have shown that it was physically impossible for respondent to have been the assailant. Nor would the preservation of the evidence here have been a burden upon the police. There obviously was refrigeration available, as the preservation of the swab indicates, and the items of clothing likely would not tax available storage space. Considered in the context of the entire trial, the failure of the prosecution to preserve this evidence deprived respondent of a fair trial. It still remains "a fundamental value determination of our society that it is far worse to convict an innocent man than to let a guilty man go free." In re Winship, 397 U. S. 358, 372 (1970) (concurring opinion). The evidence in this case was far from conclusive, and the possibility that the evidence denied to respondent would have exonerated him was not remote. The result is that he was denied a fair trial by the actions of the State, and consequently was denied due process of law. Because the Court's opinion improperly limits the scope of due process, and ignores its proper focus in a futile pursuit of a bright-line rule,[10] I dissent. NOTES [*] In this case, the Arizona Court of Appeals relied on its earlier decision in State v. Escalante, 153 Ariz. 55, 734 P. 2d 597 (1986), holding that " `when identity is an issue at trial and the police permit destruction of evidence that could eliminate a defendant as the perpetrator, such loss is material to the defense and is a denial of due process.' " 153 Ariz. 50, 54, 734 P. 2d 592, 596 (1986), quoting Escalante, supra, at 61, 734 P. 2d, at 603 (emphasis added). The reasoning in Escalante and the instant case mark a sharp departure from Trombetta in two respects. First, Trombetta speaks of evidence whose exculpatory value is "apparent." 467 U. S., at 489. The possibility that the semen samples could have exculpated respondent if preserved or tested is not enough to satisfy the standard of constitutional materiality in Trombetta. Second, we made clear in Trombetta that the exculpatory value of the evidence must be apparent "before the evidence was destroyed." Ibid. (emphasis added). Here, respondent has not shown that the police knew the semen samples would have exculpated him when they failed to perform certain tests or to refrigerate the boy's clothing; this evidence was simply an avenue of investigation that might have led in any number of directions. The presence or absence of bad faith by the police for purposes of the Due Process Clause must necessarily turn on the police's knowledge of the exculpatory value of the evidence at the time it was lost or destroyed. Cf. Napue v. Illinois, 360 U. S. 264, 269 (1959). [1] The Court's discussion in Trombetta also noted other cases: In Napue v. Illinois, 360 U. S. 264 (1959), the prosecution failed to inform the defense and the trial court that one of its witnesses had testified falsely that he had not been promised favorable treatment in return for testifying. The Court noted that a conviction obtained by the knowing use of such testimony must fall, and suggested that the conviction is invalid even when the perjured testimony is " `not the result of guile or a desire to prejudice. . . for its impact was the same, preventing, as it did, a trial that could in any real sense be termed fair.' " Id., at 270, quoting People v. Savvides, 1 N. Y. 2d 554, 557, 136 N. E. 2d 853, 854-855 (1956). In Giglio v. United States, 405 U. S. 150 (1972), the Court required a federal prosecutor to reveal a promise of nonprosecution if a witness testified, holding that "whether the nondisclosure was a result of negligence or design, it is the responsibility of the prosecutor." Id., at 154. The good faith of the prosecutor thus was irrelevant for purposes of due process. And in Roviaro v. United States, 353 U. S. 53 (1957), the Court held that in some cases the Government must disclose to the defense the identity of a confidential informant. There was no discussion of any requirement of bad faith. [2] The Agurs Court went on to note that the standard to be applied in considering the harm suffered by the defendant was different from the standard applied when new evidence is discovered by a neutral source after trial. The prosecutor is "the `servant of the law, the twofold aim of which is that guilt shall not escape or innocence suffer.' " 427 U. S., at 111, quoting Berger v. United States, 295 U. S. 78, 88 (1935). Holding the prosecution to a higher standard is necessary, lest the "special significance to the prosecutor's obligation to serve the cause of justice" be lost. 427 U. S., at 111. [3] Nor does United States v. Valenzuela-Bernal, 458 U. S. 858 (1982), provide support for the majority's "bad faith" requirement. In that case a defendant was deprived of certain testimony at his trial when the Government deported potential witnesses after determining that they possessed no material evidence relevant to the criminal trial. These deportations were not the result of malice or negligence, but were carried out pursuant to immigration policy. Id., at 863-866. Consideration of the Government's motive was only the first step in the due process inquiry. Because the Government acted in good faith, the defendant was required to make "a plausible showing" that "the evidence lost would be both material and favorable to the defense." Id., at 873. In Valenzuela-Bernal, the defendant was not able to meet that burden. Under the majority's "bad faith" test, the defendant would have no opportunity to try. [4] In Killian, the notes in question related to witnesses' statements, were used to prepare receipts which the witnesses then signed, and were destroyed in accord with usual practice. 368 U. S., at 242. Had it not been the usual practice of the agents to destroy their notes, or if no reports had been prepared from those notes before they were destroyed, a different question, closer to the one the Court decides today, would have been presented. [5] The cases relied upon by the majority for the proposition that bad faith is necessary to show a due process violation, United States v. Marion, 404 U. S. 307 (1971), and United States v. Lovasco, 431 U. S. 783 (1977), concerned claims that preindictment delay violated due process. The harm caused by such delay is certainly more speculative than that caused by the deprivation of material exculpatory evidence, and in such cases statutes of limitations, not the Due Process Clause, provide the primary protection for defendants' interests. Those cases are a shaky foundation for the radical step taken by the Court today. [6] We noted in California v. Trombetta, 467 U. S. 479, 486 (1984): "The absence of doctrinal development in this area reflects, in part, the difficulty of developing rules to deal with evidence destroyed through prosecutorial neglect or oversight. Whenever potentially exculpatory evidence is permanently lost, courts face the treacherous task of divining the import of materials whose contents are unknown and, very often, disputed." While the inquiry is a difficult one, I do not read Trombetta to say, nor do I believe, that it is impossible. Respect for constitutional rights demands that the inquiry be made. [7] There is no need in this case to discuss whether the police have a duty to test evidence, or whether due process requires that police testing be on the "cutting edge" of technology. But uncertainty as to these questions only highlights the importance of preserving evidence, so that the defense has the opportunity at least to use whatever scientifically recognized tests are available. That is all that is at issue in this case. [8] This Court "has recognized the inherently suspect qualities of eyewitness identification evidence." Watkins v. Sowders, 449 U. S. 341, 350 (1981) (BRENNAN, J., dissenting). Such evidence is "notoriously unreliable," ibid.; see United States v. Wade, 388 U. S. 218, 228 (1967); Manson v. Brathwaite, 432 U. S. 98, 111-112 (1977), and has distinct impacts on juries. "All the evidence points rather strikingly to the conclusion that there is almost nothing more convincing than a live human being who takes the stand, points a finger at the defendant, and says, `That's the one!' " E. Loftus, Eyewitness Testimony 19 (1979). Studies show that children are more likely to make mistaken identifications than are adults, especially when they have been encouraged by adults. See generally Cohen & Harnick, The Susceptibility of Child Witnesses to Suggestion, 4 Law and Human Behavior 201 (1980). Other studies show another element of possible relevance in this case: "Crossracial identifications are much less likely to be accurate than same race identifications." Rahaim & Brodsky, Empirical Evidence versus Common Sense: Juror and Lawyer Knowledge of Eyewitness Accuracy, 7 Law and Psych. Rev. 1, 2 (1982). These authorities suggest that eyewitness testimony alone, in the absence of corroboration, is to be viewed with some suspicion. [9] The victim testified that the car had a loud muffler, that country music was playing on its radio, and that the car was started using a key. Respondent and others testified that his car was inoperative on the night of the incident, that when it was working it ran quietly, that the radio did not work, and that the car could be started only by using a screwdriver. The police did not check any of this before disposing of the car. See 153 Ariz. 50, 51-52, 734 P. 2d 592, 593-594 (App. 1986). [10] Even under the standard articulated by the majority the proper resolution of this case should be a remand to consider whether the police did act in good faith. The Arizona Court of Appeals did not state in its opinion that there was no bad faith on the part of the police. Rather, it held that the proper standard to be applied was a consideration of whether the failure to preserve the evidence deprived respondent of a fair trial, and that, as a result, its holding did "not imply any bad faith on the part of the state." Id., at 54, 734 P. 2d, at 596. But there certainly is a sufficient basis on this record for a finding that the police acted in bad faith. The destruction of respondent's car by the police (which in itself may serve on remand as an alternative ground for finding a constitutional violation, see id., at 55, 734 P. 2d, at 597 (question left open)) certainly suggests that the police may have conducted their investigation with an improper animus. Although the majority provides no guidance as to how a lack of good faith is to be determined, or just how egregious police action must be, the police actions in this case raise a colorable claim of bad faith. If the Arizona courts on remand should determine that the failure to refrigerate the clothing was part of an overall investigation marred by bad faith, then, even under the majority's test, the conviction should be overturned.