*671OPINION
This is an appeal from the denial of a petition for writ of habeas corpus .challenging petitioner Emanuel Sistrunk’s state court conviction for rape. We must decide whether Sistrunk has made a sufficient showing of his actual innocence to pass through the Schlup gateway. We conclude that he has not; therefore, we affirm the district court.
I. BACKGROUND
Sistrunk was convicted in Oregon state court in 1986 for the rape of an eleven-year-old girl. The victim testified that she knew Sistrunk, who had assaulted her once before and threatened to kill her family if she told. When she saw him shortly after school got out on the day in question, she tried to run away, but he caught up with her, grabbed her by the arm and took her into a garage, where he laid his coat on a square oil pan on the floor and proceeded to rape her. Sistrunk then slapped the victim on the face with a five-dollar bill, showed her back to school, and warned that he would kill her and her family if she told anyone. The victim bought some flowers for her mother with the five dollars she had been given, took the after-school activity bus home, and eventually told her mother what had happened. Two days later, she was examined by an emergency room doctor who noted that she had an abrasion in her vaginal area, some vaginal discharge, and complained of having a burning sensation when she urinated.
Responding to defense counsel’s questions to describe Sistrunk’s penis, the victim stated that his penis “had bumps on it.” Sistrunk suffers from neurofibromato-sis, which causes pigmented spots and bumps on the skin. Sistrunk’s counsel moved the court to admit into evidence a photograph of Sistrunk’s erect penis to show the jury that it had no bumps. The motion was denied. The court did, however, permit Sistrunk’s girlfriend to testify to the absence of bumps on Sistrunk’s penis.
At trial, the state’s main expert witness, Dr. Jan Bays, testified that studies on child abuse allegations proved that “it is very, very rare that a child lies about sex abuse.” She specifically referred to a yet-to-be-published study that concluded that “2 percent of the [child abuse] cases were false allegations. Those were not — they were never younger children.” She summarized the study as follows: “So, the criteria that came out of the study were, if the child comes forward with the story, themselves [sic] it is the truth. If the child is younger than a teenager, then it is the truth.”
' The jury found Sistrunk guilty and the court sentenced him to 30 years’ imprisonment, with a 15-year minimum. His conviction was affirmed on direct appeal. State v. Sistrunk, 85 Or.App. 558, 737 P.2d 978 (Or.Ct.App.), rev. denied, 303 Or. 590, 739 P.2d 570 (Or.1987). His first petition for state post-conviction relief was denied, and the denial was affirmed on appeal by the Oregon Court of Appeals. Sistrunk v. Wright, 99 Or.App. 488, 782 P.2d 958(Or.Ct.App.1989). Sistrunk did riot seek review of the denial of his first petition by the Oregon Supreme Court. He then filed a second petition for state post-conviction relief, which was also denied. Again, the denial was affirmed on appeal. Sistrunk v. Zenon, 140 Or.App. 644, 917 P.2d 77 (Or.Ct.App.), rev. denied, 323 Or. 483, 918 P.2d 847 (Or.1996). On November 5, 1996, Sistrunk filed a federal petition for writ of habeas corpus, which was later amended. In his amended petition, Sistrunk challenged the legality of his conviction on the grounds of ineffective assis*672tance of counsel, insufficiency of the evidence, and prosecutorial misconduct.
Sistrunk’s ineffective assistance claim attacked the competence of both his trial and appellate counsel. He alleged that his trial counsel was ineffective for failing to object to Dr. Bays’ testimony, thereby allowing Dr. Bays impermissibly to vouch for the credibility of the victim and mislead the jury regarding the scientific study on which she relied. He also alleged that his appellate 'counsel failed to raise the issue of the trial court’s refusal to admit a photograph of his penis to demonstrate to the jury that it did not have bumps.1 Sistrunk concedes that, because he did not seek review of his ineffective assistance of counsel claim by the Oregon Supreme Court in his first state post-conviction proceeding, it is procedurally defaulted. He argues, however, that this default should be excused because he is actually innocent and therefore entitled to present the merits of his underlying claims.
The district court denied Sistrunk’s petition. In her findings and recommendations, the magistrate judge specifically stated that, although it was true that Sis-trunk’s counsel had committed errors durr ing the state court,proceedings, Sistrunk could not demonstrate that Dr. Bays’ impermissible vouching for the victim’s testimony or the evidence of the smooth condition of Sistrunk’s penis constituted the type of “new evidence” necessary to pass through the Schlup gateway to reach Sis-trunk’s claims of constitutional error. On de novo review, the district court adopted the magistrate judge’s findings and recommendations, and denied the writ. Sistrunk v. Armenakis, 1999 WL 717214 (D.Or.1999).
This timely appeal followed. We have jurisdiction under 28 U.S.C. §§ 1291 and 2253.2 A divided panel of this court affirmed the district court. Sistrunk v. Armenakis, 271 F.3d 1174, 1181 (9th Cir. 2001). We subsequently granted rehearing en banc. Sistrunk v. Armenakis, 279 F.3d 1063(9th Cir.2002).
II. STANDARD OF REVIEW
We review de novo the denial of a petition for writ of habeas corpus. Karis v. Calderon, 283 F.3d 1117, 1126(9th Cir. 2002).
III. DISCUSSION
The sole issue on appeal is whether Sistrunk has presented new evidence sufficient to allow him to avoid the procedural default of his ineffective assistance of counsel claim under Schlup v. Delo, 513 U.S. 298, 115 S.Ct. 851, 130 L.Ed.2d 808 (1995).3 In order to present otherwise procedurally barred claims to a federal habeas court, a petitioner must come forward with sufficient proof of his actual innocence to bring him “within ‘the *673narrow class of cases ... implicating a fundamental miscarriage of justice.’ ” Id. at 314-15, 115 S.Ct. 851(quoting McCleskey v. Zant, 499 U.S. 467, 494, 111 S.Ct. 1454, 113 L.Ed.2d 517 (1991)). Actual innocence can be shown when a petitioner “presents evidence of innocence so strong that a court cannot have confidence in the outcome of the trial unless the court is also satisfied that the trial was free of non-harmless constitutional error.” Schlup, 513 U.S. at 316, 115 S.Ct. 851. In contrast to the standard for “actual innocence” applicable to claims of substantive constitutional error, see Herrera v. Collins, 506 U.S. 390, 113 S.Ct. 853, 122 L.Ed.2d 203 (1993), a petitioner’s “claim of innocence [under Schlup] is ... ‘not itself a constitutional claim, but instead a gateway through which a habeas petitioner must pass to have his otherwise barred constitutional claim considered on the merits.’ ” Schlup, 513 U.S. at 315, 115 S.Ct. 851(quoting Herrera, 506 U.S. at 404, 113 S.Ct. 853).
In order to pass through the Schlup gateway, a “petitioner must show that it is more likely than not that no reasonable juror would have convicted him in the light of the new evidence.” Id. at 327, 115 S.Ct. 851 (internal citation omitted). In assessing whether a petitioner has met this standard, the reviewing tribunal may “consider the' probative force of relevant evidence that was either excluded or unavailable at trial.” Id. at 327-28, 115 S.Ct. 851. We held in Carriger v. Stewart, 132 F.3d 463, 478-79(9th Cir.1997) (en banc), that where post-conviction evidence casts doubt on the conviction by undercutting the reliability of the proof of guilt, but not by affirmatively proving innocence, that can be enough to pass through the Schlup gateway to allow consideration of otherwise barred claims.
Sistrunk contends that he has met the Schlup standard by offering previously unavailable evidence demonstrating the falsity of Dr. Bays’ testimony and previously excluded evidence regarding the true condition of his penis that contradicts the victim’s assertion that it was bumpy.4 According to Sistrunk, this new evidence so undermines confidence in the outcome of his trial that this court must allow him to pass through the Schlup gateway and have his constitutional claims considered. He emphasizes that because his trial was not “untainted by constitutional error,” we aré not permitted to accord the conviction “the same degree of respect as one ... that is the product of an error-free trial” in analyzing his showing of procedural innocence. Schlup, 513 U.S) at 316-17, 115 S.Ct. 851. Accepting our obligation to view the underlying state trial court decision with a skeptical eye, we nevertheless conclude that the new evidence is insufficient to meet the Schlup standard.
*674Sistrunk’s main contention is that evidence of the false and biased nature of Dr. Bays’ testimony regarding child sexual abuse allegations is sufficient for him to pass through the Schlup gateway. Specifically, Sistrunk contends that Dr. Bays lied when she asserted that sexual abuse allegations made by pre-teen children are truthful, since the then-unpublished study upon which she relied in her trial testimony did not support her statements. The study constitutes new evidence, according to Sistrunk, because it was published the year' following Dr. Bays’ testimony and was therefore unavailable’ for impeachment purposes at trial. Sistrunk further argues that Dr. Bays was biased, as revealed by a law review article she published after the trial that explains ways to admit studies assessing the credibility of sexually abused children.5
The study in question, published in the Journal of Interpersonal Violence in 1987, evaluated accounts of sexual abuse of children from two different data sets. Phase 1 of the study found that of 576 reports of child sexual abuse made to the Denver Department of Social Services. in 1983, “[m]ost reports were rehable accounts (70%), but a small proportion appeared to be fictitious (8%).” David P.H. Jones & J. Melbourne McGraw, Reliable and Fictitious Accounts of Sexual Abuse in Children, 2 J. Interpersonal Violence 27, 27 (1987). Phase 2 of the study evaluated a sample of 21 fictitious cases of child abuse identified at the Kempe National Center between 1983 and 1985. Id. át 36. During that same time period, 696 reliable cases of child abuse were also seen. Id. Of the fictitious accounts in Phase 1, there were a total of eight made by five children — four of the children weré “disturbed female teenagers who had been sexually victimized by an adult in the past,” while “[t]he fifth child was 4 years old and produced an account with his mother that appeared fictitious ... Id. at 30. Of the 21 fictitious reports from Phase 2, five were made by girls between the ages of three and nine.6 Id.
Thus, Dr. Bays’ trial testimony contained a number of inaccurate statements. Although she asserted that the study “found that 2 percent of the cases were false allegations,” the figure for Phase 1 was actually eight percent. Dr. Bays stated that the false allegations were never made by younger children — “the younger children always told the truth,” she testified — although, in fact, one report (out of 576) in Phase 1 was made by a pre-teen child, while five (out of 717) in Phase 2 were pre-teen allegations. Dr. Bays’ overall assessment of the study was that if a pre-teen child comes forward with the story, “then it is the truth.” However, there were five pre-teen children from Phase 2 who came forward by themselves with false allegations (the pre-teen accuser from Phase 1 was accompanied by a parent). Thus, although Dr. Bays truthfully stated, based on the study, that “it is very, very rare that a child lies about sex abuse,” her statement that young children never make false allegations was belied by the evidence — six of the 1,293 allegations studied *675(.005%) in Phases 1 and 2 were .falsely made by pre-teen children.
Although Dr. Bays’ testimony was clearly inaccurate, it is certainly not clear that it constituted a “lie,” especially in the legal sense of the word.7 What is clear is that Dr. Bays overstated the study’s conclusions and failed to heed the author’s admonition that the study was “uncontrolled” and that “it would be improper to offer conclusions without reference to what degree of doubt exists.” Id. at 38, 43. Nevertheless, we cannot conclude that the discrepancies in Dr. Bays’ testimony so undermine confidence in the proof of Sis-trunk’s guilt that it is probable that no reasonable jury would have convicted him. Had the jury been confronted with the true facts about the study — that less than one percent of the sexual abuse allegations under consideration were falsely made by pre-teenage children — it is unlikely that it would have altered the outcome. In fact, a precise discussion of the study’s findings would have still supported the victim’s case.8
Similarly, Sistrunk’s allegations of bias do not provide much support for his Schlup claim. Tex begin with, his assertions of bias- are thin: by Sistrunk’s own account, Dr. Bays’ co-authored law review article simply stated that, despite courts’ rejection of expert testimony that comments directly on the credibility of individual children, experts could nevertheless rebut attacks on credibility by drawing from the clinical and scientific literature for the conclusion that fabricated allegations are rare. Meyers et al., Expert Testimony in Child Sex Abuse Litigation,' 68 Neb. L.Rev. 1, 121-27 (1989). We simply fail to see how proposing legitimate methods for introducing evidence of scientific studies on child abuse allegations constitutes bias.
Moreover, even if we were to accept Sistrunk’s argument that the evidence of falsity and bias thoroughly compromises the bolstering effect of Dr. Bays’ testimony, it does nothing to discredit the victim’s account of the rape, which was central to establishing guilt. Indeed, at trial, the victim presented a coherent narrative of the events that was subjected to cross examination. The victim recounted her abduction, described the garage where the rape took place, and repeated how she used the five dollars Sistrunk gave her to *676buy flowers for her mother. The victim’s prompt disclosure of the rape and the medical evidence of her vaginal abrasion further supported her account. The jury, of course, saw and heard the victim testify and therefore was afforded the first-hand opportunity to judge her credibility.
Sistrunk contends that evidence of the condition of his penis, excluded by the trial court, would have undermined the victim’s credibility, leaving the jury with such doubts about the prosecution’s case that it would not have found him guilty beyond a reasonable doubt. Specifically, a medical exam conducted at the pehitentiary after trial confirmed that his penis was free from neurofibromatosis tumors and therefore not “bumpy” as the victim alleged. This evidence, however, although it has some impeachment value, is far from conclusive. For one, the absence of neurofib-romas on Sistrunk’s penis is not inconsistent with the victim’s description — that is, it would not be unusual for an eleven-year-old girl to describe a normal adult penis as bumpy. In addition, a young girl being raped by a man with neurofibromas conspicuously spotting his body might not necessarily notice the absence of such tumors on his penis. Thus, her attribution of bumpiness to his penis might simply reflect her dominant recollection of his physical characteristics during the traumatic event. Given that the victim’s description of Sistrunk’s penis was such a small part of the overall evidence against him, we cannot say that the victim’s impeachment on this, point would have constituted evidence of innocence so strong as to undermine confidence in the conviction.
This conclusion is reinforced by the fact that the evidence supporting Sistrunk’s alibi defense is not particularly strong. According to the victim, the rape took place between 3:00 p.m. and 4:00 p.m. Sistrunk testified that during this time frame he walked from a community center to a near-by grocery store, and then to the welfare office with some friends. Sis-trunk’s own witnesses, however, could not corroborate his alibi. Ransom Eddings, with whom Sistrunk met at the community center, testified that he saw Sistrunk in the afternoon from approximately 1:10 p.m. to 2:00 p.m. Samuella Organ, the grocery store owner, testified that Sistrunk left her store between 1:30 p.m. and 2:00 p.m. and did not return until approximately 6:30 p.m'. None of the defense’s other witnesses could account for Sistrunk’s whereabouts during the critical time in question. While a petitioner’s alibi does not have to be air-tight, in light of the overall record in this case, the absence of corroboration and the inconsistencies in Sistrunk’s story militate against concluding that Sistrunk’s newly-presented evidence is sufficient to allow him to pass through the Schlup gateway.9
At bottom, Sistrunk’s Schlup claim relies on an attempt to discredit prosecution witnesses, rather than affirmatively presenting new exculpatory evidence. Sis-trunk’s evidence does not purport to show that he did not rape the victim, only that Dr. Bays should not have been allowed to vouch for the girl’s testimony and that the victim should have been further impeached by photographic evidence that Sistrunk’s penis had no bumps. Under Carriger, impeachment evidence, by itself, can demonstrate actual innocence, where it gives rise to “sufficient doubt about the validity of [the] conviction.” Carriger, 132 F.3d at 478; see also Schlup, 513 U.S. at 330, 115 S.Ct. 851 (“[UJnder the gateway standard *677..., the newly presented evidence may indeed call into question the credibility of the witnesses presented at trial.”).
This case, however, is distinguishable from Carriger, in which we held that the petitioner had met the Schlup standard where he had presented a sworn statement from a third party confessing to the crime for which the petitioner was convicted. Carriger, 132 F.3d at 478. The confession “accurately described details about the crime and the crime scene that only a participant could have known” and explained why the petitioner’s fingerprint was found on adhesive tape used to bind the hands of the murder victim. Id. There was additional evidence indicating that the third party had boasted of committing the crime; described how he disposed of the evidence and maintained the lie that the petitioner had committed the murder; recounted the gruesome nature of the murder; and had a long history of violence, lying to the police, and blaming others for his crimes. Id. at 478-79. Although this evidence did not affirmatively show the petitioner’s actual innocence under the more stringent Herrera standard, we held that it was nevertheless sufficient to allow the petitioner to pass through the Schlup gateway. Id. at 477-79.
In contrast, the evidence that Sistrunk has provided is not nearly as compelling as the detailed third-party confession in Car-riger, which undermined the validity of the prosecution’s entire case by showing that the critical witness against the petitioner was the culpable party. Here, the newly-presented. evidence of Dr. Bays’ supposedly false testimony and Sistrunk’s non-bumpy penis certainly provides a basis for some degree of impeachment of the prosecution’s main witnesses. It does not, however, fundamentally call into question the reliability of Sistrunk’s conviction. If the jury had been presented with an accurate reading of the article cited by Dr. Bays in her testimony, it. would - still have been likely to convict, given the extreme infrequency of false sexual abuse allegations by pre-teen children. Even if the allegedly false and biased nature of Dr. Bays’ testimony would have devastated her credibility, it would not have cast doubt on the first-hand account of the victim, who positively identified Sistrunk as the rapist in open court. Further, given that the victim’s description of Sistrunk’s penis as bumpy was the non-clinical characterization of an eleven-year-old child, evidence that Sistrunk’s penis did not have , neuro-fibromas would not have had much persuasive force with respect to whether the victim was testifying falsely about the rape. Therefore, even if all of the evidence Sistrunk now proffers had been introduced at the trial, we are unconvinced, in light of the victim’s testimony and Sis-trunk’s shaky alibi, that “it is more likely than not that no reasonable juror would have convicted him.” Schlup, 513 U.S. at 327, 115 S.Ct. 851.
IY. CONCLUSION
Based on the foregoing, we conclude that Sistrunk has failed to present sufficient evidence of actual innocence to permit him to pass through the Schlup gateway and to argue the merits of his procedurally barred claims. The judgment of the district court is, therefore, AFFIRMED.