View PDF Version

No. 08-559


In the Supreme Court of the United States






Solicitor General
Counsel of Record
Assistant Attorney General
Deputy Solicitor General
Assistant to the Solicitor
Department of Justice
Washington, D.C. 20530-0001
(202) 514-2217


The United States will address the following ques tion:

Whether a court, in evaluating the sufficiency of the evidence under Jackson v. Virginia, 443 U.S. 307 (1979), may disregard evidence presented to the jury if the court concludes that additional evidence submitted after trial demonstrates that the trial evidence was inaccurate or unreliable.

In the Supreme Court of the United States

No. 08-559








This case presents the question whether a federal court sitting in habeas review of a state conviction may, in evaluating a sufficiency-of-the-evidence claim under Jackson v. Virginia, 443 U.S. 307 (1979), disregard evi dence presented to the jury if the court determines that additional evidence submitted after trial shows that the trial evidence was inaccurate or unreliable. Because criminal defendants regularly challenge federal convic tions under Jackson, which establishes a due process standard that applies to all prosecutions, and because the Double Jeopardy Clause precludes retrial after a finding of evidentiary insufficiency, resolution of that question will have substantial implications for direct and collateral review of federal convictions. The United

States therefore has a significant interest in the Court's disposition of this case.


In the early morning hours of January 29, 1994, nine- year-old Jane Doe (a pseudonym) was raped in her home. Pet. App. 3a & n.3. Circumstantial evidence linked respondent to the crime, and forensic analysis of semen recovered from Jane's underwear showed that the semen's deoxyribonucleic acid (DNA) matched re spondent's DNA profile. After a Nevada state jury con victed respondent of the sexual assault and the convic tion was affirmed on appeal, the District Court for the District of Nevada granted respondent habeas relief. Id. at 31a-62a. The Ninth Circuit affirmed, concluding that evidence submitted after trial showed that the State's DNA expert gave inaccurate and misleading trial testimony and, without that testimony, the remaining trial evidence was constitutionally insufficient under Jackson. Id. at 3a, 10a, 13a-21a.

1. DNA is a complex molecule that contains the ge netic code of organisms. See David H. Kaye & George F. Sensabaugh, Jr., Federal Judicial Center, Reference Guide on DNA Evidence, in Reference Manual on Sci entific Evidence 485, 491, 508 (2d ed. 2000) (DNA Guide) < sciman09.pdf/$file/sciman09.pdf>. Most human DNA is contained in chromosomes located within the nuclei of most cells. Ibid. Those cells normally have 23 pairs of chromosomes, which are formed by joining 23 chromo somes from each parent. Ibid.

Roughly 99.9% of the chromosomal DNA in every human is identical. DNA Guide 491-492. The remaining 0.1% reflects genetic variability unique to each individ ual (except for identical twins). Ibid. A chromosomal location at which the DNA sequence will vary between individuals is known as a polymorphic locus. There is a discrete number of potential, alternative genetic se quences (known as alleles) found at any such locus, and, at those loci, individuals inherit one allele from each bio logical parent. Id. at 492-494 & n.36, 508.

Forensic DNA testing compares the genetic material in biological samples by examining the alleles in those samples at specific, strategically selected polymorphic loci. DNA Guide 493; see, e.g., id. at 494, 500-502, 506- 507 (discussing Restriction Fragment Length Polymor phism (RFLP) analysis); id. at 493-494 & n.32, 497-500, 506-507 (discussing Polymerase Chain Reaction (PCR) analysis); National Inst. of Justice, U.S. Dep't of Justice, The Future of Forensic DNA Testing 14-19 (2000) <>. Nor mally, DNA profiles developed from crime-scene evi dence are compared to the profiles of suspect(s) and others (e.g., the victim or a spouse) who may have left biological evidence at the crime scene.

If any allele at any tested loci of an individual's DNA clearly does not match the corresponding allele from the crime-scene evidence, the individual can be excluded as the source of that evidence. DNA Guide 516. Alterna tively, if all the tested loci show identical alleles, the genetic profiles will match and the individual will be a possible source of the crime-scene DNA. Ibid. Such a forensic "match," however, reflects only that the individ ual's DNA matches the crime-scene DNA at the tested loci. Ibid. It does not necessarily show that the individ ual's DNA is the same as the crime-scene DNA, and the possibility remains that another individual also could have a matching profile for the tested loci. Ibid.; see The Future of Forensic DNA Testing 20.1

Forensic DNA experts traditionally explain the sig nificance of matching a suspect's DNA profile to a crime-scene DNA sample by calculating and testifying to the sample's "random-match probability," i.e., the probability that an individual would have the same DNA profile as the crime-scene sample if that individual were selected at random from a relevant population. The Fu ture of Forensic DNA Testing 20; DNA Guide 524-525. Once the alleles from crime-scene DNA are identified, the frequency of each allele's occurrence in the popula tion (based on large-scale population studies) may be used to estimate the probability that an individual se lected at random from that population would have a ge notype with the same alleles at all the tested loci. Id. at 525; The Future of Forensic DNA Testing 20, 23; cf. David H. Kaye et al., The New Wigmore: A Treatise on Evidence § 12.4.1, at 457 n.5 (2004) (It is "no longer * * * seriously disputed" that "the computations asso ciated with most DNA evidence satisfy Daubert [v. Mer rell Dow Pharmaceuticals, Inc., 509 U.S. 579 (1993)]." (citing DNA Guide)). That random-match probability, if sufficiently small, will show that it is implausible that the profiles of the suspect and the crime-scene DNA have matched based on mere coincidence. Id. at 457; DNA Guide 524-525.2

2. On the evening of January 28, 1994, Pamela Doe, the mother of Jane Doe, went to a bar with her neigh bors, Trent and Raquel Brown. Pet. App. 4a, 70a. While they were out, Jane and her four-year-old sister went to Trent and Raquel's trailer home to babysit Raquel's children. Id. at 4a, 70a. Raquel and Trent returned home around 7:30 p.m. and Raquel took Jane and her sister back to their own trailer around 9:30 p.m. Ibid. Because Pamela remained at the bar and the girls' step- father was at work, the girls were at home alone. Ibid.

Jane telephoned local bars to find her mother, even tually reaching respondent, the brother of Trent Brown. Pet. App. 4a, 70a. Respondent told Jane he would find her mother and deliver the message that the girls were home safely. Respondent did so and subsequently had drinks with Pamela at another bar. Id. at 4a, 71a. Pam ela last saw respondent between 11:00 p.m. and mid night. One bartender said respondent left by 12:20 a.m., while another put him in the bar at 1:30 a.m. Ibid.

Sometime in the early morning of January 29, Jane was sexually assaulted in her bedroom. Between mid night and 1:00 a.m., Jane called her mother at the bar and said that a man who had been at the trailer looking for Pamela had hurt her. Pet. App. 4a-5a, 71a. Pamela hurried home to find her daughter covered in blood from the waist down, with bruises on her neck and scratches on her face. Ibid. Hospital personnel determined that Jane had been penetrated both vaginally and anally. Jane lost about 15% of her blood, required vaginal re construction, was hospitalized for 12 days, and acquired post-traumatic stress disorder. Id. at 5a, 72a-73a.

Jane described her attacker as hatless with blonde or sandy-colored hair and perhaps a small moustache. Pet. App. 5a. She stated that he wore dark jeans, a black jacket with a zipper, a western-type shirt, boots, and a watch which scraped her face. Ibid. Jane indicated that she had bitten her assailant's hands and that he had smelled "awful," like "beer or puke or something." Ibid.

That night, respondent was wearing a cowboy hat, dark jeans, boots, and a black jacket with an orange and yellow logo on the back. Pet. App. 5a. Two witnesses testified that they saw a man wearing such an outfit (but with a green emblem on the back) near Jane's trailer around 1:05 a.m. Respondent lived on the same street as Jane, approximately ten trailers away. Id. at 5a-6a, 76a; J.A. 603.

Respondent stated that he consumed 20 alcoholic drinks that night and vomited multiple times, soiling his pants and shirt. He denied wearing a watch. Pet. App. 6a, 73a-74a. When respondent arrived at his trailer-at 1:32 am, according to his second brother and roommate, Travis Brown-respondent washed his clothes. Respon dent claimed he did this because he was leaving on a trip the next day and his other clothes were packed. Ibid. A police officer who arrived at 5:00 a.m. to question re spondent did not see bite marks on his hands and found no blood on respondent or his boots. Although Jane speculated that her assailant must have turned off her night light, a fingerprint on that light did not belong to respondent, nor were his fingerprints found elsewhere in Jane's trailer. Ibid.

When the police asked Jane who the attacker re minded her of, she vacillated between respondent and his brother Trent. Pet. App. 6a-7a, 75a. Days later, Jane saw a TV report on respondent's arrest and identi fied him as her assailant. She also stated that the man she had seen on TV had sent her flowers. The flowers, however, had come from Raquel and Trent. Ibid. When the police showed Jane pictures that included a likeness of respondent, Jane could not identify him as her at tacker. Id. at 7a, 75a-76a.

3. a. On February 7, 1994, respondent was charged with sexual assault and attempted murder. Pet. App. 76a-77a, 187a. In February and June 1994, Renee Ro mero, a forensic scientist at the Washoe County Crime Lab, completed reports concluding that Polymerase- Chain-Reaction (PCR)-based testing revealed that re spondent's alleles at six loci matched those of DNA from semen collected from Jane's underwear. J.A. 746-751.

The state trial court provided respondent with fund ing for his own DNA expert, and respondent's counsel retained Jennifer Mihalovich to assist in respondent's defense. Pet. App. 78a; J.A. 1184, 1424-1425. Respon dent's counsel had numerous conversations with Mihal ovich, who advised him that the Washoe County Crime Lab was "fully capable of performing accurate DNA testing," that Romero's expertise qualified her as a DNA expert, and that she found Romero's testing procedures appropriate and was satisfied with the accuracy of Romero's results. J.A. 1185, 1425-1428, 1492.

On July 6, 1994, Romero testified as a DNA ex pert at respondent's preliminary hearing. Pet. App. 131a; 7/6/94 Tr. 151-180. On cross-examination, Romero agreed that Restriction Fragment Length Polymor phism (RFLP) analysis had greater discriminatory power than the PCR-based tests previously completed but that she did not believe she had enough evidence to perform RFLP testing, which required larger samples of non-degraded DNA. Id. at 172-176.

Romero subsequently located additional semen on Jane's underwear and performed RFLP testing. J.A. 432, 446, 464-465. Her report, dated September 15, 1994, explained that respondent's DNA matched the semen sample at five (additional) tested loci and that "the frequency of this matching pattern is rarer than 1 in 3,000,000 in the Caucasian, Black, and Hispanic popu lations." J.A. 752-753.

Repondent's expert, Michalovich, reviewed Romero's RFLP results and was satisfied they were accurate. J.A. 1185, 1438-1439, 1492. When respondent's counsel asked whether additional time to perform independent testing would be useful, Michalovich advised that "any and all further DNA testing would do nothing but fur ther provide devastating in[crimin]atory evidence against [respondent]." J.A. 1439. Respondent's counsel later explained that he "[a]bsolutely [did] not" want DNA testing of yet-untested vaginal fluid collected from Jane after her rape. J.A. 1430-1431. He also did not ask Mihalovich for written reports, which could be "ex tremely damaging" and "haunt [respondent's] file" if they were later subject to discovery on collateral review. J.A. 1469.

b. Respondent's jury trial began on September 27, 1994, and ended four days later. Pet. App. 132a; J.A. 60- 687. The State presented numerous witnesses, including Romero. See J.A. 411-474 (Romero testimony). The tri al court recognized Romero "as an expert in DNA analy sis and comparison," and admitted into evidence Ro mero's RFLP forensic report (J.A. 752-753) and her re sume (J.A. 755-760), all without objection. J.A. 414-415, 463, 1186.

Romero described the process of DNA analysis with RFLP testing, J.A. 415-424, and explained that this test ing showed that respondent's genetic profile as deter mined by five DNA probes matched the semen recov ered from Jane Doe's underwear. J.A. 425-439. Romero further explained that population studies reveal the fre quency of occurrence for each DNA band (allele) identi fied in the test and that scientists "us[e] a statistical cal culation" developed by the National Research Council as a "conservative method" of estimating the likelihood of a match in the population. J.A. 437; see J.A. 422-423, 428. Romero testified that she employed this statistical method and determined that roughly one in three mil lion people in the general population would have a DNA profile matching the semen evidence. J.A. 437-440.

The State's prosecutor asked Romero if there were a way of expressing the one-in-three-million statistic as "the likelihood that the DNA found in the panties is the same as the DNA found in [respondent's] blood." J.A. 458. Romero eventually acceded to the prosecutor's re quest to express this likelihood as a percentage and con cluded that "[i]t would be 99.999[9]67 percent." Ibid.3 The prosecutor had Romero use a blackboard to sub tract that percentage from 100% and admitted her writ ten calculation into evidence. J.A. 459-460. The prose cutor then asked Romero if it "would * * * be fair to say" based on her subtraction that "the chances that the DNA found in the panties * * * is not [respondent's] would be .000033?" J.A. 460. Romero agreed that the statement was "not inaccurate." J.A. 461-462. The judge asked whether "it's the same math just expressed differently," and Romero responded: "Yes. Exactly." Ibid.

Romero admitted in cross-examination that the sta tistical probability would change if the suspect had brothers and that, "in this case," based on National Re search Council data, the probability that a brother would have the same DNA profile as respondent would be "one in 6,500." J.A. 468-469.

Respondent did not call his own DNA expert at trial, and the jury convicted respondent of felony sexual as sault of a child under the age of 14 causing substantial bodily harm, felony sexual assault of a child under the age of 14, and abuse or neglect of a child under the age of 18. Pet. App. 78a; J.A. 783-786.

c. Before sentencing, the trial court granted respon dent's request to have his own DNA expert test vaginal fluid recovered from Jane Doe. Pet. App. 133a; J.A. 797- 800. After conducting a PCR-based test (with less dis criminatory power than RFLP testing), Mihalovich and a colleague submitted a report (J.A. 910-922) concluding that respondent's DNA matched the forensic evidence at the tested alleles and that only "approximately 0.01% [one out of 10,000] of the Caucasian population" shared those genetic markers. J.A. 917 (brackets in original). At sentencing, the court stated that "there is no doubt in my mind whatsoever of [respondent's] guilt" and that respondent's own DNA testing was "further proof of what we already knew from the first DNA testing, that this vicious crime was committed by [respondent]." J.A. 797, 800.

d. The Nevada Supreme Court concluded that the trial evidence, including the State's DNA evidence, was sufficient to permit "a jury, acting reasonably," to find respondent guilty beyond a reasonable doubt. Pet. App. 82a-83a. The court, however, reversed respon dent's abuse-and-neglect conviction on double-jeopardy grounds, identified sentencing errors, and remanded for resentencing. Id. at 83a-96a. Respondent was resen tenced to two consecutive terms of life imprisonment, each with the possibility of parole after ten years. Id. at 63a-67a. Respondent's subsequent appeal was dis missed. J.A. 1077-1080.

e. In 2001, respondent filed a state habeas-corpus petition, which was denied after an evidentiary hearing. J.A. 1489-1499; J.A. 1036-1047, 1202-1488. The habeas court, inter alia, rejected petitioner's argument that his counsel was ineffective, and the Nevada Supreme Court affirmed. J.A. 1492-1506.

4. a. In December 2003, respondent filed a federal habeas petition in district court. J.A. 1. In 2006, the court granted respondent's request to supplement the record with a report (J.A. 1581-1584) by Dr. Lawrence Mueller, a professor of Ecology and Evolutionary Biol ogy at the University of California, Irvine. J.A. 1595- 1598.

Mueller's February 2006 report criticized two as pects of Romero's 1994 trial testimony. First, it stated that the "accepted meaning" of the statistics for a DNA match is that they "represent the chance that a single person chosen at random from the suspect population would match the evidence." J.A. 1583. Mueller opined that Romero was "not correct" in suggesting that there was only a 0.000033% likelihood of "anyone other than [respondent] matching the evidence sample" and that her error reflected "erroneous phraseology * * * so common" that it is known as "the prosecutor's fallacy." Ibid.

Mueller opined that Romero also incorrectly testified to a one-in-6500 chance that one brother would match respondent's DNA profile and that the probability was, at worst, one in 1024. J.A. 1582. Mueller stated that he had "been told" that respondent had two brothers who lived near the crime scene and two others at "a greater distance away." Ibid. Mueller opined that, if two of re spondent's brothers are considered, "the chance that one or more brothers would match the evidence is 1 in 512" under Romero's assumptions and, if all four are considered, the odds increase to one in 256. Ibid. Muel ler added that, under his own analysis, the probability of at least one of those four brothers matching the DNA profile was "1 in 66." J.A. 1583. The State did not sub mit evidence disputing Mueller's opinions.

b. The district court granted habeas relief and or dered a new trial. Pet. App. 29a-54a. It concluded that the Nevada Supreme Court failed to apply the correct legal standard in evaluating respondent's sufficiency-of- the-evidence claim, and that Romero's trial testimony must be "set[] aside" because it was "inaccurate and, therefore, unreliable." Id. at 37a-46a. Absent that testi mony, the court found the trial evidence insufficient un der Jackson and held that the Nevada Supreme Court's contrary ruling violated clearly established federal law. Id. at 46a, 54a. The court additionally held that respon dent was denied the effective assistance of counsel be cause his attorney failed to object to the admission of the State's DNA evidence, to present a defense expert at trial, and to ensure that a defense expert observed Romero's DNA testing. Id. at 46a-50a.

c. The court of appeals affirmed in a divided opinion. Pet. App. 1a-28a. The majority held, inter alia, that the Nevada Supreme Court's determination that the trial evidence was sufficient to allow a "reasonable jury" to convict was contrary to this Court's "rational"-jury stan dard in Jackson. Id. at 13a-14a. The majority further held that the state court's application of Jackson was "unreasonable" because, "in light of the Mueller Report, no rational trier of fact could have found [respondent] guilty beyond a reasonable doubt on the evidence pre sented at trial." Id. at 14a-15a; see id. at 14a-21a. The court accordingly held that respondent was denied due process under Jackson and was entitled to habeas relief without reaching respondent's ineffective-assistance claim. Id. at 3a, 21a.

The court explained that Romero's trial testimony "was later proved to be inaccurate and misleading" on federal habeas review because "[t]he Mueller Report [now] indicates that Romero's testimony was unreliable for two main reasons." Pet. App. at 3a, 15a. First, the court found that Romero's testimony concerning a 99.99967% probability that respondent's DNA was the same as the DNA in the semen sample reflected the "prosecutor's fallacy" by incorrectly "confus[ing] source probability with random match probability." Id. at 15a. Citing scholarly articles and court opinions on DNA evi dence, the court concluded that the "probability of find ing a random match can be much higher than the proba bility of matching one individual" to the crime and, for that reason, Romero's statements were "misleading" and probably caused respondent to be "convicted based on the jury's consideration of false, but highly persua sive, evidence." Id. at 16a-17a. Second, the court found that Romero "inaccurately minimized the likelihood that [respondent's] DNA would match one of his four broth ers' DNA" because Mueller estimated that the odds were 1/66, roughly ten times greater than Romero's 1/6500 figure. Id. at 17a-18a.

The court then reasoned that a "federal court on ha beas [review] may exclude evidence admitted in the state court" if that evidence has "rendered [the] trial so fundamentally unfair as to violate federal due process," and that the "[a]dmission of [Romero's] unreliable testi mony most certainly rendered the trial fundamentally unfair." Pet. App. 18a-19a (citation omitted). "After ex cluding Romero's testimony," the court agreed with the district court that "the remaining evidence [viewed] in the light most favorable to the prosecution" was insuffi cient under Jackson to permit "any rational trier of fact to believe that [respondent] was the assailant beyond a reasonable doubt." Id. at 19a-21a; see id. at 14a (evi dence insufficient "once the unreliable DNA testimony is excluded"). The court ordered respondent to be re tried within 180 days or released. Id. at 21a.

Judge O'Scannlain dissented. Pet. App. 21a-28a. He concluded that the Nevada Supreme Court complied with Jackson, reasoning that a Jackson claim is based on "the record evidence adduced at the trial" and can not be resolved "by imagining a different state trial in which evidence actually presented would have been ex cluded-especially not on the basis of reports added to the record during federal habeas review." Id. at 22a, 26a (quoting Jackson, 443 U.S. at 324). Even accepting Mueller's post-trial analysis, Judge O'Scannlian ex plained, the DNA and circumstantial evidence was suffi cient to support respondent's conviction. Id. at 26a-28a.


The evidence at trial, which included powerful DNA evidence linking respondent to Jane Doe's rape, was constitutionally sufficient for a rational jury to convict respondent beyond a reasonable doubt. The Ninth Cir cuit's contrary holding based on post-trial evidence fun damentally misconstrues Jackson v. Virginia, 443 U.S. 307 (1979).

Jackson concluded that due process requires that factfinders rationally apply the proof-beyond-a-reason able-doubt standard to the facts in evidence. It there fore held that a conviction violates due process if "no rational trier of fact could have found proof of guilt be yond a reasonable doubt" based "upon the record evi dence adduced at the trial." 443 U.S. at 324. Jackson's evidentiary threshold does not ask whether the factfind er was correct, but whether its verdict was rational based on the evidence before it. Reviewing courts ac cordingly must (1) restrict their Jackson review only to the "record evidence adduced at the trial" and (2) exam ine "all of the evidence" in the light most favorable to the prosecution. Id . at 319, 324. If that evidence is in sufficient, the conviction must be reversed and double jeopardy precludes retrial.

Romero's expert testimony, with the other trial evi dence, far exceeded Jackson's threshold. The Ninth Circuit did not conclude otherwise. It instead found, based on Mueller's report and other information submit ted after trial, that Romero's testimony was partially inaccurate or unreliable and, excluding all of her testi mony, that the remaining evidence was insufficient. This analytical approach is foreclosed by Lockhart v. Nelson, 488 U.S. 33 (1988), which held that all the trial evidence must be considered when evaluating a suffi ciency-of-the-evidence claim, even evidence conclusively shown after trial to be incorrect. The jury did not have access to such subsequently submitted information. Nor could the jury be thought to have possessed such infor mation on its own. The jury thus considered a different evidentiary record than the one on which the court of appeals conducted sufficiency analysis. By starting with evidence adduced at trial, adding Mueller's post-trial report, subtracting evidence that the court found unreli able, and weighing the remaining evidence for suffi ciency, the Ninth Circuit charted a fundamentally un sound course, untethered from Jackson's rationale and holding.

The Ninth Circuit also erred in holding that Romero's testimony must be excluded because its admis sion rendered respondent's trial fundamentally unfair. The State provided respondent a DNA expert to assist in his defense, and respondent had a full opportunity to present evidence and cross-examine Romero effectively to expose any unreliable aspects of her testimony. Re spondent's failure to exercise his full spectrum of trial rights does not render his trial fundamentally unfair. At best, respondent may have an ineffective assistance of counsel claim, which, unlike his Jackson claim, would not preclude respondent's retrial.



A. Jackson's Rationality Test Turns On The Evidence That Was Before The Factfinder

In Jackson v. Virginia, 443 U.S. 307 (1979), this Court concluded that the due-process requirement that criminal convictions rest on proof beyond a reasonable doubt entails that "the factfinder will rationally apply that standard to the facts in evidence." Id. at 316-317. A "reasonable doubt," the Court explained, must be "one based upon 'reason,'" and, if a judge or jury convicts based on evidence that would permit "no rational trier of fact [to] find guilt beyond a reasonable doubt," the resulting conviction "cannot constitutionally stand." Id. at 317-318 & n.9. Jackson accordingly held that a crimi nal conviction violates due process if "no rational trier of fact could have found proof of guilt beyond a reasonable doubt" based "upon the record evidence adduced at the trial." Id. at 324 & n.16; see id. at 319 n.12.

Jackson accordingly establishes a "constitutional minimum" of evidentiary sufficiency, 433 U.S. at 319 & nn.12-13, that defines the threshold at which "the trier of fact has power to make a finding of guilt." Schlup v. Delo, 513 U.S. 298, 330 (1995). That inquiry "does not focus on whether the trier of fact made the correct guilt or innocence determination, but rather whether it made a rational decision to convict or acquit." Herrera v. Col lins, 506 U.S. 390, 402 (1993).4 If "no rational factfinder could have voted to convict the defendant" on the evi dence before it, the conviction must be reversed and the Double Jeopardy Clause precludes retrial. Tibbs v. Florida, 457 U.S. 31, 41, 45 (1982) (citing Jackson and Burks v. United States, 437 U.S. 1 (1978)); see id. at 51 (White, J., dissenting) ("[R]etrial is foreclosed by the Double Jeopardy Clause if the evidence fails to satisfy the Jackson standard.").5

At least two principles flow from Jackson's focus on the factfinder's ability to "rational[ly]" convict on the "record evidence adduced at trial," 443 U.S. at 317, 324. First, Jackson claims must be evaluated only on the evidence that was actually presented to the factfinder. The "critical inquiry" under Jackson is "whether the record evidence could reasonably support a finding of guilt beyond a reasonable doubt." Id. at 318. A court conducting that inquiry must limit itself to the evidence that the factfinder actually could have considered-i.e., the evidence adduced at trial-because a decision may be fully rational when made on one set of information yet entirely irrational on another. This Court has ac cordingly made clear that the "eviden[tiary] review au thorized by Jackson is limited to 'record evidence'" and "does not extend to non-record evidence, including newly discovered evidence" submitted after trial. Herrera, 506 U.S. at 402 (quoting Jackson, 443 U.S. at 318); see House v. Bell, 547 U.S. 518, 538 (2006) (distin guishing Jackson claims from gateway actual-innocence claims "involv[ing] evidence the trial jury did not have before it").

Second, a reviewing court must "review all of the evidence" admitted at trial and evaluate that evidence for sufficiency "in the light most favorable to the prose cution." Jackson, 443 U.S. at 319. That deferential standard preserves "the factfinder's role as weigher of the evidence" and impinges on its authority "only to the extent necessary to guarantee the fundamental protec tion of due process of law." Ibid. By focusing on whether "any rational trier of fact could have found the essential elements of the crime beyond a reasonable doubt," ibid., Jackson ensures that the factfinder pos sessed the "the power * * * to reach its conclusion" based on the record that was before it, Schlup, 513 U.S. at 330, without second-guessing the factfinder's conclu sions.

Jackson's minimal requirement of rationality does not turn on whether the factfinder "made the cor rect guilt * * * determination," Herrera, 506 U.S. at 402, and it does not authorize a reviewing court to "make its own subjective determination of guilt or innocence." Jackson, 443 U.S. at 320 n.13. Instead, a court "faced with a record of historical facts that supports conflicting inferences" must "presume * * * that the trier of fact resolved any such conflicts in favor of the prosecution" and "defer to that resolution." Id. at 326. Reversals for evidentiary insufficiency are therefore "confined to cases where the prosecution's failure is clear." Tibbs, 457 U.S. at 41 (quoting Burks, 437 U.S. at 17).

B. The Evidence Presented At Trial Was Sufficient To Find Respondent Guilty Beyond A Reasonable Doubt

1. The evidence adduced at respondent's trial far exceeds Jackson's constitutional threshold for suffi ciency. We assume arguendo that the trial evidence linking respondent to Jane Doe's rape would be insuffi cient in the absence of any DNA evidence connecting respondent to the crime. Cf. Pet. App. 3a & n.1, 10a, 19a (citing purported concession by petitioners); J.A. 1176. But cf. Pet. Br. 35 & n.6 (disputing concession).6 On that assumption, Romero's uncontradicted, expert testimony provided three evidentiary bases which, when combined with the other evidence at trial, would permit a rational jury to find respondent guilty beyond a reasonable doubt.

First, Romero testified that only one in about three million individuals would exhibit a DNA profile match ing that of the semen retrieved from Jane's underwear, and Romero's corresponding report (J.A. 752-753) was admitted in evidence. J.A. 437-440, 463. The trial court qualified Romero as an "expert in DNA analysis" in the jury's presence without opposition, and Romero's re sume (J.A. 755-760) was in evidence. J.A. 414-415. Re spondent introduced no expert testimony at trial that might have contradicted Romero's analysis, and even Mueller's post-trial report (J.A. 1581-1584) does not dis pute that there is only a one-in-three-million probability of a random DNA match for the semen sample. The match of respondent's DNA with this exceedingly rare genetic profile would, with the other evidence in the case, permit a rational juror to find respondent guilty beyond a reasonable doubt.

Second, Romero's testimony indicated that there was a 99.999[9]67% likelihood that the DNA in the semen sample belonged to respondent or, stated differently, only a 0.000033% chance that the DNA was not from re spondent. J.A. 458-462. That testimony went unchal lenged at trial, and the conclusion of the court of appeals that this aspect of Romero's testimony was "highly per suasive," Pet. App. 17a, effectively acknowledges that the testimony is sufficient to support a rational convic tion.

Finally, Romero testified that there was only a 1-in- 6500 likelihood that one of respondent's brothers would have the same DNA profile. J.A. 468-469. That uncon tradicted evidence would permit a rational jury to con clude beyond a reasonable doubt that respondent, rather than a brother, committed the rape. Especially given the trial evidence indicating that respondent's brothers (Trent and Travis) had alibis at the time of the rape, see Pet. Br. 29, a rational jury could have eliminated the brothers as assailants.7

2. Although the court of appeals found that Muell er's report showed that the latter two aspects of Ro mero's testimony were inaccurate or unreliable, Pet. App. 15a-18a, the jury could not have based its verdict on the reasoning in Mueller's report because it was writ ten 12 years after respondent's trial. Expert evidence that was not presented at trial cannot undermine the reliability of the trial evidence for purposes of applying the Jackson standard. The only question under Jackson is whether the jury could have rationally convicted "upon the record evidence adduced at the trial," 443 U.S. at 324, and "newly discovered evidence" is irrele vant, Herrera, 506 U.S. at 402. Thus, unless other trial evidence could demonstrate overwhelmingly that Ro mero's testimony was flawed, a rational jury properly could rely upon it, even if in hindsight and with the ben efit of a newly prepared expert report, its reliability can now be questioned.

That conclusion follows from Lockhart v. Nelson, 488 U.S. 33 (1988), which held that all evidence admitted at trial (including evidence subsequently shown to be incor rect) must be considered in a court's sufficiency calculus. In Nelson, the State sought to prove the applicability of a recidivist sentencing enhancement beyond a reason able doubt by submitting into evidence four prior felony convictions. Id. at 34-36. The jury found that Nelson had four such convictions and imposed an enhanced sen tence. Id. at 36. The State subsequently determined that one of the convictions had been pardoned. Id. at 36-37. A federal district court then invalidated the sen tencing enhancement on habeas review and held that double jeopardy precluded the State from resentencing Nelson. Id. at 37. The court of appeals affirmed, con cluding that it was error to have introduced the par doned conviction into evidence and, without it, the State failed to submit constitutionally sufficient evidence. Ibid.

This Court reversed, holding that the proper review of a double jeopardy claim based on "insufficiency of the evidence" must "consider all of the evidence admitted by the trial court." Nelson, 488 U.S. at 41 (citing Burks, 437 U.S. at 16-17) (emphasis added).8 The Court ex plained that, although there would be "insufficient evi dence to support a judgment of conviction" without evi dence of the (pardoned) offense, there "clearly" was "enough to support the sentence" "with that evidence" because the "jury had before them certified copies of four prior felony convictions" which, on the record as it then existed, were "sufficient to support a verdict of en hancement." Id. at 40. The Court recognized that Nel son had not submitted evidence to the jury "to prove that the conviction had become a nullity" and observed that, if he had, the State presumably would have had "an opportunity to offer [other] evidence * * * to support the habitual offender charge." Id. at 42.9

Nelson is fatal to respondent's Jackson claim. By requiring that "all" trial evidence be considered, Nelson underscores that the purpose of a sufficiency-of-the-evi dence claim is to test the rationality of a verdict from the perspective of the factfinder at the time the verdict was rendered. If the trial evidence is later shown to be incorrect by information outside the trial record, that showing will not alter the rationality of the verdict based on that evidence, before its reliability had been under mined. All evidence before the factfinder must be weighed, not to determine whether its verdict was "cor rect" (Herrera, 506 U.S. at 402), but to assess whether that determination was rational. Jackson, it might be said, does not permit Monday-morning quarterbacking.

3. In cases in which a jury convicts after hearing expert testimony, a court reviewing a sufficiency-of-the- evidence claim may consider that testimony and con clude that the conviction must be reversed if no rational juror could have found guilt beyond a reasonable doubt on the evidence as a whole. But the court must be care ful not to evaluate the evidence using specialized knowl edge obtained from sources outside the trial record if the prototypical juror would not have had such informa tion in rendering the verdict.10

Expert testimony, by its very nature, "often will rest 'upon an experience confessedly foreign in kind to [the jury's] own.'" Kumho Tire Co. v. Carmichael, 526 U.S. 137, 149 (1999) (brackets in original; citation omitted). Experts must testify in federal court to "scientific, tech nical, or other specialized knowledge," Fed. R. Evid. 702, and a central criterion for admissibility is whether the expertise would assist laypersons. Thus, a court must conduct "the common sense inquiry whether the untrained layman would be qualified to determine intel ligently and to the best possible degree the particular issue without enlightenment from those having special ized understanding of the subject." Id. advisory comm. note (citation omitted); see 4 Weinstein's Federal Evi dence § 702.03[1], at 702-34 & n.1 (2d ed. 2009) (expert testimony is admissible if it "assists" the factfinder by addressing "matters beyond the understanding of the average person"); cf. Salem v. United States Lines Co., 370 U.S. 31, 35 (1962). That principle is embraced by Nevada state courts, which admit expert testimony "to provide the trier of fact a resource for ascertaining truth in relevant areas outside the ken of ordinary laity." Townsend v. State, 734 P.2d 705, 708 (Nev. 1987); see Hallmark v. Eldridge, 189 P.3d 646, 650 (Nev. 2008) (Nev. Rev. Stat. § 50.275 tracks Rule 702); Banks ex rel. Banks v. Sunrise Hosp., 102 P.3d 52, 60 (Nev. 2004) (ex perts may testify "on matters outside the average per son's common understanding").

The advanced probability and genetic principles that persuaded the court of appeals that Romero's testimony was inaccurate, Pet. App. 15a-18a, involve matters that a typical juror would not have considered in rendering a verdict. Neither "judges [n]or juries are experts in statistics," Metro-North Commuter R.R. v. Buckley, 521 U.S. 424, 435 (1997), and the "analysis of * * * DNA found at a crime scene" surely involves "factual matters outside the jurors' knowledge." United States v. Schef fer, 523 U.S. 303, 313 (1998) (plurality opinion). Indeed, the court of appeals and district court recognized that the relevant DNA science and "statistical calculation[s]" are "highly complex and difficult to understand even for the well[-]educated and patient student." Pet. App. 17a n.5, 39a.

The prosecutor's fallacy, for instance, is a colloquial name for an advanced probability concept known as the "fallacy of the transposed conditional." The New Wig more § 12.4.1.b(1), at 463.11 Such complex principles require specialized knowledge that most educated lay persons-and, indeed, most lawyers-lack. Cf. id. at 464-465 ("Although the logical fallacy in transposing the conditional seems obvious enough-after it has been pointed out-the case reports are replete with it."). The court of appeals' own reliance on Mueller's report, judi cial opinions, scholarly articles, and a "complicated for mula known as Bayes's Theorem," Pet. App. 15a-17a, illustrates that its evaluation of the trial evidence was premised on knowledge well beyond what a typical layperson would possess-and therefore, well beyond what anyone could expect a rational jury in this case to have considered.

Similarly, without specialized knowledge of genetics, DNA, and probability theory, a typical juror would not have a sound reason to reject the accuracy of Romero's uncontradicted assessment of the likelihood that respon dent's brothers would share respondent's DNA profile. Cf. J.A. 468-472. The court of appeals' repeated obser vation that the evidence was unreliable "in light of the Mueller Report," Pet. App. 8a, 15a, itself reflects that a typical juror would not have identified inaccura cies in Romero's testimony. Again, then, the informa tion deemed critical by the court of appeals cannot be thought to have entered into the analysis of the jury.

4. In short, the Ninth Circuit's analytical approach to the Jackson inquiry is fundamentally unsound. The court purported to perform a Jackson analysis by start ing with the trial evidence, adding Mueller's post-trial report, and subtracting the evidence that the court itself found unreliable based on that report. By "weigh[ing] the sufficiency of the remaining evidence," Pet. App. 19a (emphasis added), the court completely untethered its decision from Jackson's underlying rationale. By adding to and subtracting from the information that the jury actually considered, the court's analysis no longer answered whether the jury rationally could have ren dered its verdict. To resolve that question, courts must base their inquiry only on the evidence actually "ad duced at the trial" and "review all of the evidence" so admitted. Jackson, 443 U.S. at 319, 324; see pp. 18-19, supra.

C. Due Process Concerns Do Not Warrant The Ninth Cir cuit's Approach to Evidentiary Insufficiency

The Ninth Circuit appears to have justified its depar ture from a proper Jackson analysis with its conclusion that due process required the exclusion of Romero's ex pert testimony. The court reasoned that the "[a]dmis sion of this unreliable testimony * * * violated [respon dent's] due process rights" by "render[ing] the trial fun damentally unfair," and the district court therefore "did not err in excluding it" from its Jackson analysis. Pet. App. 18a-19a. That analysis is flawed for at least two reasons. First, even if Romero's testimony was improp erly admitted, Nelson makes clear that such evidence must be considered in the sufficiency calculus. See pp. 22-23, supra.12 But more fundamentally, the admission of Romero's testimony did not render respondent's trial fundamentally unfair. Criminal defendants have numer ous constitutional rights that guarantee a fair trial pro cess, and the Due Process Clauses do not generally im pose thresholds of evidentiary reliability.

1. Outside the specific guarantees of the Bill of Rights, this Court has "defined the category of infrac tions that violate 'fundamental unfairness' very nar rowly," Dowling v. United States 493 U.S. 342, 352 (1990), and it has explained that questions concerning the reliability of evidence, such as those raised by the court of appeals in this case, are "governed by the evi dentiary laws of the forum," "not by the Due Process Clause." Colorado v. Connelly, 479 U.S. 157, 167 (1986); see Spencer v. Texas, 385 U.S. 554, 563-564 (1967) ("[I]t has never been thought that [cases concerning the due process guarantee of fundamental fairness] establish this Court as a rule-making organ for the promulgation of state rules of criminal procedure."); Kansas v. Ventris, No. 07-1356 (Apr. 29, 2009), slip op. 7 n.* (re jecting exclusionary rule for uncorroborated statements of jailhouse snitches on the theory that they constitute "inherently unreliable" testimony, reasoning that "it is the province of the jury to weigh the credibility of com peting witnesses"). In particular, the Court has rejected the contention that it is "fundamentally unfair" to admit evidence alleged to be "inherently unreliable" where, as here, the defendant "had the opportunity to refute it." Dowling, 493 U.S. at 353.

That conclusion reflects that, while "[t]he Constitu tion guarantees a fair trial through the Due Process Clauses," it "defines the basic elements of a fair trial largely through the several provisions of the Sixth Amendment." Strickland v. Washington, 466 U.S. 668, 684-685 (1994). That Amendment confers upon defen dants rights to the effective assistance of counsel, id. at 687, to compulsory process, Taylor v. Illinois, 484 U.S. 400, 408-409 (1988), and to confrontation and an oppor tunity for effective cross-examination, Delaware v. Fensterer, 474 U.S. 15, 19-20 (1985) (per curiam). Thus, "[v]igorous cross-examination, presentation of contrary evidence, and careful instruction on the burden of proof are the traditional and appropriate means of attacking shaky but admissible evidence" from expert witnesses. Daubert v. Merrill Dow Pharms., Inc., 509 U.S. 579, 596 (1993). The Constitution provides defendants with "a full and fair opportunity to probe and expose * * * in firmities" in the testimony of prosecution experts through the adversary process, Fensterer, 474 U.S. at 20, 22, thereby providing a procedural guarantee, "not that evidence be reliable," but that "reliability be as sessed * * * in the crucible of cross-examination." Crawford v. Washington, 541 U.S. 36, 61 (2004).

Due process plays a role in supplementing these trial rights, but it does so by providing, in certain instances, the means for a defendant to seek to rebut the prosecu tion's experts at trial. See Ake v. Oklahoma, 470 U.S. 68, 74 (1985) (due process requires the State to provide access to a psychiatrist to assist a defendant who shows that his sanity is likely to be an important trial issue and that he cannot otherwise afford one). The Due Process Clauses do not provide a general judicial check on the reliability of evidence. Rather, the Court's cases make clear that due process aids in constructing a procedural framework from which juries, not judges, are to "resolve differences in opinion" between experts "on the basis of the evidence offered by each party." Id. at 81.13

Here, the State provided respondent with a DNA expert for trial; respondent was entitled to subpoena witnesses; and his counsel had a full opportunity to cross-examine Romero effectively. Respondent's ability to call his own DNA expert gave him ample means "to suggest to the jury that [Romero] had relied on a theory which the defense expert considered baseless," and, combined with respondent's opportunity for cross-exam ination, it provided respondent with a fair process to contest any trial testimony "marred by forgetfulness, confusion, or evasion." Fensterer, 474 U.S. at 20, 22.

Respondent's own failure to exercise the full spec trum of his procedural rights-by failing to object to Romero's testimony, probe more fully Romero's now- disputed statements during cross-examination, or pres ent a defense expert at trial-cannot form the basis of a subsequent due process claim. The admission of Romero's testimony did not render the trial fundamen tally unfair, and the jury was entitled to rely upon it. At best, respondent may have a Strickland claim -an issue the court of appeals found unnecessary to resolve. Pet. App. 21a.14

The Due Process Clauses impose meaningful obliga tions on prosecutors that facilitate a fair, adversarial process and constrain improper uses of expert testi mony. For example, due process is violated if prosecu tors intentionally present perjured testimony by ex perts, Mooney v. Holohan, 294 U.S. 103, 112 (1935), fail to correct testimony that they know to be false, Napue v. Illinois, 360 U.S. 264, 269 (1959), or fail to disclose to defendants either material exculpatory or impeachment evidence. Brady v. Maryland, 373 U.S. 83, 87 (1963); Giglio v. United States, 405 U.S. 150, 153-154 (1972). Those restrictions are not at issue where, as here, ex pert testimony may include good-faith misstatements that are not identified before the jury's verdict. In any event, prosecutorial errors of this kind do not implicate concerns of evidentiary insufficiency and would permit retrial.

2. The consequences of the court of appeals' error are significant. Reversals for evidentiary insufficiency trigger double jeopardy protections that preclude re trial. See p. 17, supra. Allowing defendants to relitigate their convictions based on newly submitted evidence, and prohibiting the State from retrying these individu als if a court finds that new evidence casts sufficient doubt on the evidence at trial, would impose serious so cial costs. Such an expansion of Jackson also would in vite collateral litigation to revisit jury verdicts and thus would undermine the jury's central role in resolving the core issues at criminal trials.15

Allowing defendants to use newly discovered mate rial to launch post-trial attacks on the reliability of trial evidence improperly benefits defendants who fail to sub mit such evidence when the prosecution could have ei ther corrected any evidentiary inaccuracies at trial or submitted additional evidence to sustain its burden of proof. For example, had respondent challenged Ro mero's testimony concerning the likelihood of respon dent's brothers having a matching DNA profile, the State might have corrected any error, challenged the opinions of respondent's expert, and submitted addi tional evidence showing that respondent's brothers could not have committed the offense.16 Indeed, respon dent's trial counsel avoided disclosing to the State his theories of alternative suspects precisely because he knew the State could "exclude all or any other possible suspects by the simple expedient of testing their DNA." J.A. 1188, 1435. Had respondent focused on his brothers as suspects during trial, "the trial judge would presum ably have allowed the prosecutor an opportunity to offer evidence" in that regard. Nelson, 488 U.S. at 42. This Court should not countenance an approach to eviden tiary sufficiency that allows defendants in this way to evade the normal trial process.


The judgment of the court of appeals should be re versed and the case remanded for further proceedings.

Respectfully submitted.

Solicitor General
Assistant Attorney General
Deputy Solicitor General
Assistant to the Solicitor

MAY 2009

1 Forensic scientists do not develop full DNA sequences because of a variety of practical and technological limitations.

2 When the DNA profile of crime-scene evidence is based on a sufficient number of loci with sufficiently rare alleles, the probability of a random match can be small enough that an expert can testify to a reasonable degree of scientific certainty that matching an individual's profile to the forensic sample will show that the individual is, in fact, the source of the forensic sample. See The Future of Forensic DNA Test ing 25-26; The New Wigmore § 12.5.3, at 503-505. Analytical techniques currently in use often enable source attribution in this manner.

3 The transcript's omission of a "9" appears to be erroneous. Cf. J.A. 754.

4 Jackson's focus on the jury's power to convict is "wholly unrelated to the question of how rationally the verdict was actually reached." 443 U.S. at 319 n.13. It therefore "does not require scrutiny of the rea soning process actually used by the factfinder-if known." Ibid.

5 Accord, e.g., Bean v. Calderon, 163 F.3d 1073, 1086 (9th Cir. 1998), cert. denied, 528 U.S. 922 (1999); Fagan v. Washington, 942 F.2d 1155, 1157 (7th Cir. 1991); United States v. Simpson, 910 F.2d 154, 159 (4th Cir. 1990).

6 It is far from clear that that assumption is correct. The court's evidentiary summary appears to resolve evidentiary inconsistencies against the State, e.g., Pet. App. 20a (relying on testimony by respon dent's brother, which the jury could disregard as biased), and ulti mately concludes that "conflicts in the evidence" were "too stark for any rational trier of fact" to convict beyond a reasonable doubt. Id. at 21a. Such reasoning ignores Jackson's teaching that the evidence must be viewed "in the light most favorable to the prosecution" by resolving "any such conflicts [in the record]" in its favor. Jackson, 443 U.S. at 319, 326; cf. Tibbs, 457 U.S. at 35-36, 45 n.21 (testimony from one eye witness "clearly satisfied" Jackson's standard even though the state court identified multiple problems with the prosecution's case, including "several factors [that] undermined [the eyewitness's] believability").

7 While the trial evidence reflects that respondent had two other brothers, J.A. 275, it does not indicate their whereabouts on the night of the rape. Cf. J.A. 1586 (post-trial evidence that other brothers, then aged 16 and 13, lived with their mother in Utah).

8 Nelson assumed, without deciding, that the Double Jeopardy Clause applies to non-capital recidivist sentencing. 488 U.S. at 37 n.6. In Monge v. California, 524 U.S. 721, 724 (1998), the Court held that it does not. Nelson's sufficiency analysis nevertheless remains valid in other settings in which a reviewing court determines, based on newly proffered evidence, that the evidence originally admitted at trial is in correct or otherwise unreliable.

9 Nelson expressed doubt that the disputed evidence was correctly described as being "inadmissible" because "[e]vidence of the disputed conviction was introduced" and the subsequent discovery of Nelson's pardon appeared most closely analogous to "newly discovered evi dence." 488 U.S. at 40 n.7. But the Court held that, even if proof of the conviction had been inadmissible, it must be considered in the suffi ciency calculus because it was, in fact, "admitted by the trial court." Id . at 41.

10 In Smith v. Mitchell, 437 F.3d 884 (9th Cir. 2006), vacated, 550 U.S. 915, and reinstated, 508 F.3d 1256, 1258 (9th Cir. 2007), for instance, no specialized knowledge of matters beyond the trial evidence was neces sary to conclude that expert testimony was insufficient to support the conviction. Shaken-Baby-Syndrome experts testified that Smith's granddaughter died from shaking but conceded the absence of the usual indicators of violent shaking and agreed that no autopsy evidence showed that a brain injury caused the child's death. Id. at 890. The experts nevertheless concluded that "death was caused by shearing or tearing of the brain stem * * * because there was no evidence in the brain itself of the cause of death." Ibid. With other evidence indicating that the crime was unlikely to have occurred, the speculative expert testimony premised on the absence of evidence was insufficient to establish guilt beyond a reasonable doubt. Ibid.

11 The prosecutor's fallacy lies in equating two distinct probabilities. The first is the probability that the individual tested has the same DNA profile as the forensic sample if the individual tested were selected at random. That probability is the random-match probability, which can be estimated from population studies and, here, is one in three million. The second probability is the probability that the individual tested was selected at random if the individual tested has the same DNA profile as the forensic sample. That probability is the likelihood that the individ ual matched purely by chance-i.e., the probability that the individual is not the actual DNA source. Though sounding linguistically similar, the actual values for those probabilities can be very different. See The New Wigmore § 12.4.1.b(1), at 463-464.

12 See also, e.g., United States v. Cruz, 363 F.3d 187, 197 (2d Cir. 2004); United States v. Miller, 146 F.3d 274, 280 (5th Cir. 1998); Simpson, 910 F.2d at 159.

13 Due process prohibits unnecessarily suggestive pre-trial police practices that corrupt a witness's ability to identify a perpetrator re liably. See Manson v. Brathwaite, 432 U.S. 98 (1977). In the absence of such a governmental practice that might distort a witness's memory, however, the Constitution does not require courts to examine in-court testimony for "indicia of reliability" because the right to cross-examina tion and other trial protections ensure a fair trial process. See United States v. Owens, 484 U.S. 554, 560-561 (1988) (declining to extend Brathwaite).

14 This case does not require the Court to resolve "whether the intro duction of an expert opinion with no basis could ever be so lacking in reliability, and so prejudicial, as to deny a defendant a fair trial." Fen sterer, 474 U.S. at 22-23 (emphasis added) (reserving question). Ro mero's overall testimony, while imperfect in parts, had a valid basis in DNA science, and respondent had a full opportunity to challenge that testimony at trial.

15 In this case, the court of appeals affirmed the district court's order that respondent be released unless retried. Pet. App. 21a, 54a. That remedy does not logically follow from a Jackson violation; an insuffi ciency holding precludes retrial. The court of appeals' mistaken under standing of Jackson cannot be recharacterized as the grant of a new trial based on due process principles, for at least two reasons. First, the court explicitly relied on Jackson sufficiency principles in granting relief. Id. at 13a-15a, 19a-21a. Second, the admission of expert evi dence claimed to be unreliable, as shown by a newly presented com peting expert report, does not itself render a trial fundamentally unfair so as to require a new trial. See pp. 28-31, supra.

16 Mueller's report is in fact premised on some questionable assump tions. For example, Mueller fails to account for the possibility that some or all of respondent's brothers may be half-brothers or unrelated by blood and fails to eliminate respondent's two brothers who appar ently were living a significant distance away in another State. J.A. 1586.