Petitioner-appellant Kenneth G. Pavel (“Pavel”) has been held for more than ten years pursuant to a state court judgment of conviction in the custody of state correctional authorities, most recently that of Respondent Appellee Melvin L. Hollins, Superintendent of the Oneida Correctional Facility. Over five years ago, on February 7, 1995, Pavel petitioned for a writ of habeas corpus under 28 U.S.C. § 2254 in the United States District Court for the Northern District of New York (Thomas J. McAvoy, then-Chief Judge). Pavel ar-. gued that the state court judgment of conviction was entered in violation of his rights under the Sixth Amendment to the United States Constitution because his trial attorney provided him with ineffective assistance. Pavel’s petition was denied, and judgment was entered accordingly.
On appeal, Pavel emphasizes that his trial attorney (1) did not prepare a defense, on the theory that the charges against Pavel would be dismissed at the close of the prosecution’s case; (2) failed to call two important fact witnesses, with the content of whose putative testimony the attorney was familiar; and (3) did not call as a witness a medical expert. In these circumstances, Pavel argues, his right to effective assistance of counsel was violated.
We agree. Accordingly, we reverse the judgment of the District Court, vacate the state court judgment of conviction, and remand the cause. On remand, the District Court shall issue a writ of habeas corpus to Mr. Pavel on the thirtieth calendar day after the issuance of our mandate unless New York State has, by that point, taken concrete and substantial steps expeditiously to retry him.
I. Background
In the spring of 1989, Pavel was arrested and accused of sexually abusing his children — Matthew Pavel (“Matthew”), then age seven, and David Pavel (“David”), then age five (collectively, “the boys”). The accusations were brought to the attention of the police by Pavel’s then-wife, Margaret Pavel (“Ms.Pavel”), with whom he was engaged in a series of marital disputes, and who had recently been denied sole custody over Matthew and David.
Upon his arrest, Pavel contacted Sanford Meltzer (“Meltzer”), the attorney who had represented him in connection with the marital disputes. Meltzer noted that there was little if any physical evidence that Pavel had sexually abused the boys in the manner that they alleged.1 Accordingly, Meltzer did not prepare a defense; instead, he planned to move to dismiss the charges against Pavel at the close of the prosecution’s presentation of its evidence, *212and was confident that the trial judge would grant the motion.2
Pavel pleaded not guilty, a bench trial commenced, and at the close of the prosecution’s evidence, Meltzer moved to dismiss the charges against Pavel. In relevant part, the motion was denied.
Meltzer, as noted, had not prepared for this eventuality. Accordingly, he had little choice — he put Pavel on the stand, and then rested without calling any other witnesses.3
On October 18, 1989, the state trial judge found Pavel guilty of some of the charged crimes and, on November 27, 1989, held a sentencing hearing, at which Pavel contended that he was innocent. Pavel was sentenced principally to two consecutive 4-12 year terms of incarceration, and was remanded to prison.
Pavel was paroled almost nine years later — on August 25, 1998. One of the conditions of his parole was that Pavel “cooperate fully” with a particular sex-offender program. As part of the program, Pavel was required to admit having committed the crimes for which he was convicted. He refused to do so, see Pavel Affidavit (executed April 20,2001) at ¶ 4 (“Because I did not commit the acts of which I was convicted, I have steadfastly refused to admit to them.”), and was therefore returned to prison on November 15, 1999. He is there now.
Before assessing Pavel’s Sixth Amendment claim, we pause briefly to provide a fuller account of the procedural history of this case, and of the evidence presented at trial.
A. Relevant Procedural History
As noted above, following a bench trial in Onondaga County Court, New York (John W. Brandt, Judge), Pavel was convicted on October 18, 1989 of sodomizing, sexually abusing, and endangering the welfare of his children. A judgment of conviction was entered, and within one calendar year, the state’s direct-appeals process had run its course: The judgment was affirmed by the Appellate Division, Fourth Department, see People v. Pavel, 163 A.D.2d 834, 558 N.Y.S.2d 369 (1990), and leave to appeal was denied by a Judge of the New York Court of Appeals, see People v. Pavel, 76 N.Y.2d 862, 560 N.Y.S.2d 1002, 561 N.E.2d 902 (1990).
On July 27, 1992, in Onondaga County Court, Pavel moved pursuant to New York Criminal Procedure Law § 440.10(1)4 to *213vacate the judgment of conviction (“the § 440.10 motion”). One of the bases of the § 440.10 motion was Pavel’s claim that Meltzer had provided him with ineffective assistance of counsel. After oral argument, at which Pavel was represented by Jeffrey S. Cook and James V. O’Gara of the New York law firm of Kelley Drye & Warren, the § 440.10 motion was denied. Leave to appeal from the denial was itself denied on April 1, 1993 by a Judge of the Appellate Division, Fourth Department.
On February 7, 1995, Pavel petitioned for a writ of habeas corpus in the District Court, arguing, inter alia, that Meltzer had provided ineffective assistance of counsel. The petition was denied as procedurally barred, and judgment was entered accordingly on August 16, 1996. Pavel then applied for a Certificate of Ap-pealability, and his application was denied in a December 23, 1996 Decision and Order entered by Judge McAvoy.
Approximately one month later, Pavel moved this Court for a Certificate of Ap-pealability. We ruled by summary order that Pavel’s claims were not procedurally barred and remanded the cause to the District Court, by mandate entered July 10, 1997, for a determination of the petition on the merits.
On June 16, 1999, Judge McAvoy again dismissed Pavel’s habeas petition, this time on the merits, and on July 20, 1999, he denied for a second time Pavel’s motion for a Certifícate of Appealability.
On October 19, 1999, Pavel moved in the Court of Appeals for another Certificate of Appealability. By order dated July 24, 2000, we granted the motion on the question of whether “appellant [Pavel] was denied effective assistance of trial counsel if, as he alleges, his counsel, assuming that his motion to dismiss would be granted, failed to prepare a defense.”
That is the question now before us.
B. Trial Evidenoe
The prosecution’s evidence at trial consisted principally of the testimony of five witnesses, which we summarize briefly here.
*214First, seven-year old Matthew testified under oath that Pavel had anally sodomized him every Saturday during the four months from December 1988 until mid-April 1989. Matthew testified also that he and David spent a week-long vacation, from Saturday, April 15, 1989 until Saturday, April 22, 1989, with Pavel, at the Florida apartment of Clothilde Pavel, Pav-el’s mother and the boys’ grandmother. Matthew testified that he was anally sodomized at least twice during that trip — at least one of the incidents, Matthew testified, took place in Clothilde Pavel’s bathroom, while she was making breakfast.5 Following the trip, Matthew testified, he and David told Ms. Pavel, their mother, about Pavel’s alleged sodomizing of them. The police were informed of the accusations; five days after returning from Florida, Matthew and David were given physical examinations, and Pavel was arrested. Matthew then spoke with a therapist named Constance McKinstry (“Ms.McKin-stry”) and the prosecutor about Pavel’s alleged sex abuse.
Second, five-year old David testified as an unsworn witness. His testimony was, as might be expected, frequently non-responsive or confused.6 But some of David’s testimony was clear; David testified unambiguously that he was anally sodomized at least three times during the 12 days before he was examined by a doctor in connection with the charges against Pavel, and that one of those episodes took place in the bathroom of the Florida apartment. At one point in his testimony, David strongly suggested that he was sodomized three times a day during the Florida trip. The prosecutor asked the following question: “You said it happened in Florida too. Do you remember how many times it happened in Florida?” David answered: “When we had lunch, breakfast and dinner.”
Third, Ms. McKinstry, who is not a licensed psychologist, testified that when she first began meeting with the boys on March 14, 1989, she believed that they suffered from “[adjustment disorder with anxious mood.” Ms. McKinstry testified that she began to suspect “the possibility” that they were being sexually abused in mid-April of 1989, when Matthew, during therapy sessions, began hitting a three and a half foot tall yellow dummy on a label located in the dummy’s genital area. Thereafter, Ms. McKinstry testified, Matthew and David talked with her during their weekly therapy sessions about Pav-el’s sexual abuse. Ms. McKinstry did not discuss the boys’ allegations with Pavel.
Fourth, Ms. Pavel testified that, on the day the boys returned from Florida, Matthew began singing songs about his “pee-ny” when she was giving the boys a bath. Matthew’s singing, Ms. Pavel testified, prompted a series of questions and a conversation in which the boys described Pav-el’s alleged sexual abuse.
Fifth, the prosecution called a medical expert, Dr. Celeste Madden (“Dr.Madden”). Dr. Madden testified that she reviewed the records of the physical examination of the boys that had been conducted on April 27, 1989, five days *215after they returned from Florida. She did not examine the boys herself. Dr. Madden observed that the records of the physical examination of Matthew showed no relevant physical abnormalities, while the records of David’s physical examination indicated only that he had “mild perianal arrythmia,” which Dr. Madden described as “discoloration of the skin, redness around the anal area.” Dr. Madden testified that it was “possible” that diarrhea could cause such “redness.” She testified also that David’s “discoloration” was “consistent” with his account of sexual abuse, and that Matthew might have been anally sodomized as he described without any physical indication of the sodomy remaining after the fact.
At the close of the prosecution’s case, Pavel testified in his own defense. His testimony consisted almost entirely of denials of the boys’ accusations. In addition, Pavel referred briefly to David’s having diarrhea during the Florida visit with Clothilde Pavel:
Q: ... Did either child become ill on the trip to Florida?
A: Matthew had a cough and sore throat while we were down in Florida and David also had diarrhea at least one day that we were down there, yes.
Q: And did that take care of itself?
A: It seemed to work itself out, yes.
II. Discussion
A. Standard of Review
We review de novo the District Court’s denial of Pavel’s petition for a writ of habeas corpus. See, e.g., Boyette v. Lefevre, 246 F.3d 76, 88 (2d Cir.2001).
Similarly, we owe no deference to the state courts’ determination that the judgment of conviction was not entered in violation of Pavel’s federal right to effective assistance of counsel. The federal habeas corpus statute was amended in 1996 by the Antiterrorism and Effective Death Penalty Act (“AEDPA”), Pub. Law No. 104-132, 110 Stat. 1214. Pursuant to AEDPA, a federal habeas court assessing a state prisoner’s federal claim must defer in some circumstances to a state court judgment that denied the claim “on the merits.” 28 U.S.C. § 2254(d) (“ § 2254(d)”). However, because the ha-beas petition now before us was filed before AEDPA’s effective date of April 24, 1996, § 2254(d) has no bearing on this case, and we review the petition under the standards dictated by “pre-AEDPA law.” See Slack v. McDaniel, 529 U.S. 473, 482, 120 S.Ct. 1595, 146 L.Ed.2d 542 (2000). Under those standards, pure questions of law are reviewed de novo, as are mixed questions of law and fact, see, e.g., Miller v. Fenton, 474 U.S. 104, 112, 106 S.Ct. 445, 88 L.Ed.2d 405 (1985); state court factual findings are “presumed ... correct” absent certain circumstances, see 28 U.S.C. § 2254(d) (1994) (enumerating these circumstances). In this case, there are no relevant state court factual findings. Accordingly, there is no basis under “pre-AEDPA law” for deferring to the state courts’ denial of Pavel’s Sixth Amendment claim.
B. Constitutional Principles
As noted above, Pavel contends that he was convicted in state court in violation of the Sixth Amendment to the United States Constitution, which vests persons charged with crimes with the right to “the Assistance of Counsel.” U.S. Const, amend. VI. The right to “Assistance of Counsel” encompasses the right to effective “Assistance of Counsel,” see, e.g., McMann v. Richardson, 397 U.S. 759, 771 n. 14, 90 S.Ct. 1441, 25 L.Ed.2d 763 (1970), and applies to the states as a component of the *216right to “due process of law” secured by the Fourteenth Amendment to the United States Constitution, see Evitts v. Lucey, 469 U.S. 387, 394, 105 S.Ct. 830, 83 L.Ed.2d 821 (1985).
To establish that he was convicted in violation of his right to effective assistance of counsel, a claimant must satisfy both prongs of the two-part test articulated in Strickland v. Washington, 466 U.S. 668, 104 S.Ct. 2052, 80 L.Ed.2d 674 (1984). This test is “rigorous,” Lindstadt v. Keane, 239 F.3d 191, 199 (2d Cir.2001), and “highly demanding,” Kimmelman v. Morrison, 477 U.S. 365, 382, 106 S.Ct. 2574, 91 L.Ed.2d 305 (1986). To satisfy it, a claimant must show both that “counsel’s performance was deficient” and “that the deficient performance prejudiced the defense.” Strickland, 466 U.S. at 687, 104 S.Ct. 2052.
As to the first prong, to determine whether an attorney’s conduct was deficient, “[t]he court must ... determine whether, in light of all the circumstances, the identified acts or omissions were outside the wide range of professionally competent assistance.” Id. at 690, 104 S.Ct. 2052 (emphasis added). As to the second prong, to establish that he was “prejudiced” by his attorney’s constitutionally deficient performance, a claimant must “show that there is a reasonable probability that, but for counsel’s unprofessional errors, the result of the proceeding would have been different.” Id. at 694, 104 S.Ct. 2052. A “reasonable probability” in this context is one that “undermine[s] confidence in the outcome.” Id.
We consider each of these prongs in turn.
C. ANALYSIS
1. Meltzer’s Performance
Pavel claims that Meltzer’s representation of him was unprofessional to the point of being constitutionally “deficient.” Strickland, 466 U.S. at 687, 104 S.Ct. 2052. We agree.
As discussed below, Meltzer’s representation of Pavel was flawed in three distinct ways. Because we conclude, that the cumulative weight of these flaws deprived Pavel of his Sixth Amendment rights, see post at 225, see, e.g., Lindstadt, 239 F.3d at 199 (holding that for Sixth Amendment purposes attorney errors must be considered “in the aggregate”), we do not consider whether some of these flaws—standing alone, or in combination with one another—could adequately support our conclusion that Meltzer’s representation of Pavel was constitutionally deficient.
a. Failure to Prepare a Defense
First, Meltzer decided not to prepare a defense for Pavel solely because he was confident that, at the close of the prosecution’s presentation of its evidence, the trial judge would grant Meltzer’s motion to dismiss the government’s charges against Pavel. That Meltzer opted not to prepare a defense based entirely on this rationale militates strongly in favor of the conclusion that his representation of Pavel was constitutionally deficient.7 See Harris v. Reed, 894 F.2d 871, 878-79 (7th Cir.1990); United States ex rel. Cosey v. Wolff, 562 F.Supp. 140, 143-46 (N.D.Ill.1983), aff'd in relevant part on opinion below by United *217States ex rel. Cosey v. Wolff, 727 F.2d 656 (7th Cir.1984), overruled on other grounds by United States v. Payne, 741 F.2d 887, 891 n. 4 (7th Cir.1984); see also Osborn v. Shillinger, 861 F.2d 612, 626 (10th Cir.1988) (holding that an attorney’s performance was deficient under Strickland’s first prong where, inter alia, the attorney failed to prepare for state court “death penalty phase” proceedings because he believed that he could convince the prosecutor not to seek the death penalty); Blake v. Kemp, 758 F.2d 523, 535 (11th Cir.1985) (Elbert P. Tuttle, Judge) (holding that an attorney’s performance was deficient under Strickland’s first prong where, inter alia, the attorney failed to prepare for state post-conviction proceedings because he believed that his client would not be convicted); cf. Jackson v. Calderon, 211 F.3d 1148, 1161-62 (9th Cir.2000); Sneed v. Smith, 670 F.2d 1348, 1354 (4th Cir.1982); Gomez v. Beto, 462 F.2d 596, 597 (5th Cir.1972).
b. Failure to Call Important Facts Witnesses
Next, Pavel notes that Meltzer decided not to call two fact witnesses — Clothilde Pavel and Dr. Ralph Berry III (“Dr.Berry”) — with whose putative testimony he was familiar before the trial began.8 Pavel contends that this decision was an extraordinarily poor one, and we agree.
Before explaining why, we pause briefly to note that Meltzer’s decision not to call particular witnesses related to trial strategy, see United States v. Luciano, 158 F.3d 655, 660 (2d Cir.1998) (“[t]he decision not to call a particular witness is typically a question of trial strategy”), and that we have been especially hesitant to disturb such “strategic” decisions.9 See, e.g., id; Trapnell v. United States, 725 F.2d 149, 155 (2d Cir.1983). See generally Strickland, 466 U.S. at 690, 104 S.Ct. 2052 (“strategic choices made after thorough investigation of law and facts relevant to plausible options are virtually unchallengeable”).
However, this hesitation to challenge a lawyer’s “strategic” decisions has no place here. Meltzer stated that he opted not to prepare a defense — and not to call any witnesses other than Pavel — solely because he believed that the motion to dis*218miss would be granted. It is apparent from this explanation that Meltzer's decision as to which witnesses to call was animated primarily by a desire to save himself labor — to avoid preparing a defense that might ultimately prove unnecessary.10 Meltzer’s decision not to call any witnesses other than Pavel was thus “strategic” in the sense that it related to a question of trial strategy — which witnesses to call. And it was “strategic” also in that it was taken by him to advance a particular goal.
That goal, however, was mainly avoiding work' — not, as it should have been, serving Pavel’s interests by providing him with reasonably effective representation. Therefore, although Meltzer’s decision was “strategic” in some senses of the word, it was not the sort of conscious, reasonably informed decision made by an attorney with an eye to benefitting his client that the federal courts have denominated “strategic” and been especially reluctant to disturb.11
*219Accordingly, the hesitation- to disturb “strategic” decisions described above has no bearing on this case. There are many ways properly to assist a client. See ante n. 9. But making important decisions with no regard for a client’s interests is not one of them. That much is perfectly clear without the benefit of hindsight. See id.
We turn now to an assessment of Melt-zer’s decision not to call either Clothilde Pavel or Dr. Berry.
(i) Clothilde Pavel
Pavel, David, Matthew, and Clothilde Pavel spent a week at the Florida apartment of Clothilde Pavel just before Pavel was arrested; the boys testified that during that week they were sodomized by Pavel.
Pavel argues that Meltzer should have called Clothilde Pavel. We agree.
The record before the District Court showed that Clothilde Pavel explained that had she been called at trial, she would have testified as follows:
On Wednesday, April 19, 1989, we drove back to my apartment in Boynton Beach at the conclusion of our trip to Disney-world. At the beginning of this 2-3 hour drive, David, the younger child, became extremely agitated because he needed to go to the bathroom. After searching along the road for a rest-stop we found an International House of Pancakes restaurant. Immediately after we pulled into the parking lot, and before [Pavel] could get out the door, David ran out of the car and into the restaurant. [Pavel] chased after him and they returned shortly thereafter at which point [Pavel] informed me that David had had trouble with his bowels and that his stomach was obviously very upset. At the time we thought nothing of it and attributed his upset stomach to the fact that he had eaten too much at breakfast and lunch and had been badly frightened on the Space Mountain ride at Disney-world.
This testimony would have been valuable to the defense. It would have tended to corroborate Pavel’s testimony that Matthew had diarrhea during the Florida trip, see ante 215, thereby bolstering the defense’s theory that there was a wholly innocent explanation (an upset stomach) for the only species of physical evidence (slight redness in David’s anal area) that so much as suggested that either of the boys had been sexually abused in Florida in the manner that they described.
More importantly, in addition to her testimony about David’s gastrointestinal difficulties, Clothilde Pavel would have testified that although Matthew stated that Pavel had sodomized him in the bathroom while she busied herself “cooking” breakfast, and David suggested that he was sexually abused in Florida during mealtimes — “lunch, breakfast and dinner” — she did not “cook” breakfast during the week in Florida. Rather, she noted, breakfast consisted of “cold cereals or other foods that required little or no preparation.” Thus, Meltzer could have argued, with some force, that it would have been difficult to sodomize Matthew or David, unnoticed, in the relatively small apartment, and in the short time that it takes to pull together a breakfast of cold cereal, or one that otherwise requires “little or no preparation.”
Clothilde Pavel stated also that it was “equally unlikely that the acts which were supposed to have occurred could have occurred while I was preparing any other meal as the two boys spent virtually all their time in front of the television set in the living room when we were not at the beach.” And in a similar vein, although Matthew testified that he was sodomized *220in the bathroom during breakfast-time, and David testified that he too was sexually abused in the bathroom, Clothilde Pavel stated that the boys were alone in the bathroom with Pavel only at night (when he helped them brush their teeth), and then only with the door open. Indeed, Clothilde Pavel stated she was “quite certain that Kenneth [Pavel] had no opportunity to abuse the boys at any time during their visit with me in Florida as the four of us literally spent all of our time together.” 12
If she had given this testimony, Cloth-ilde Pavel would have served as something of an alibi witness for Pavel. Her testimony would have damaged the credibility of the boys’ assertions that Pavel had sexually abused them as charged in the indictment. It would have suggested that Pavel could not have abused the boys in Florida, as they claimed, because Pavel was not in the right place (in the bathroom with the boys) at the right time (during breakfast, or even during other meals).
“[T]he Constitution does not oblige counsel to present each and every witness that is suggested to him.” United States v. Balzano, 916 F.2d 1273, 1294 (7th Cir.1990). But in light of the testimony that Clothilde Pavel would have offered — and in light of the fact that Meltzer knew of this testimony before trial — Meltzer’s decision not to call Clothilde Pavel was an error of extraordinary magnitude.
First, there is simply no suggestion in the record that there was any reason not to put Clothilde Pavel on the stand, and “an attorney’s failure to present available exculpatory evidence is ordinarily deficient, unless some cogent tactical or other consideration justified it.” Griffin v. Warden, 970 F.2d 1355, 1358 (4th Cir.1992) (internal quotation marks omitted) (collecting cases); cf. Brecheen v. Reynolds, 41 F.3d 1343, 1368 (10th Cir.1994) (similar).
And second, Meltzer’s decision not to put Clothilde Pavel on the stand was based on an inadequate investigation.13 As noted above, Meltzer was familiar with the basic contours of Clothilde Pavel’s testimony before the trial — presumably because he had spoken about the matter with Pavel. But Meltzer never followed-up on what he learned of Clothilde Pavel’s putative testimony with Clothilde Pavel herself: Melt-zer never contacted her with regard to her putative testimony, and never inquired into whether she might be willing to testify on Pavel’s behalf.
Meltzer should have done both things. It should have been obvious to Meltzer *221from, the outset that Pavel’s trial — like many sex abuse trials — was going to be a so-called “credibility contest,” in which the word of the defendant’s alleged victims (David and Matthew) would be set against the word of the defendant (Pavel), and in which there would be no substantial circumstantial evidence. See generally Brian L. Schwalb, Note, Child Abuse Trials and the Confrontation of Traumatized Witnesses: Defining “Confrontation” to Protect Both Children and Defendants, 26 Harv. C.R.-C.L. L.Rev. 185, 186 (1991) (“often it is only the child and the [alleged] abuser who know what transpired”). In such cases, it should be perfectly obvious that it will almost always be useful for defense counsel to speak before trial with readily-available fact witnesses whose noncumulative testimony would directly corroborate the defense’s theory of important disputes. Accordingly, when Meltzer learned before trial that Clothilde Pavel might well be such a witness, he should have taken affirmative steps to discuss the case with her.14 Meltzer should not have relied only on what he had learned from Pavel. Cf. United States v. Moore, 554 F.2d 1086, 1093 (D.C.Cir.1976) (stating that “counsel’s anticipation of what a potential witness would say does not excuse the failure to find out,” in the context of an attorney who failed to contact a particular witness because the attorney was confident as to what the witness would say). Rather, at the very least, Meltzer should have directly contacted Clothilde Pavel — to learn more about the testimony that she would have offered, and to determine whether she might serve as a suitable witness based, inter alia, on his assessment of her credibility. But Meltzer simply did not contact Clothilde Pavel. Indeed, there is no indication in the record that Meltzer conducted any substantial, affirmative investigation into Clothilde Pavel’s potential testimony.15
*222In sum, Meltzer knew, before the trial, that Clothilde Pavel would have provided useful testimony with regard to David’s stomach troubles and, more importantly, would have provided non-cumulative testimony that was “akin to an alibi.” Lindstadt, 239 F.3d at 200-01. In light of this knowledge, Meltzer’s decision not to call Clothilde Pavel was flawed because it was not based on either a plausible strategic calculus or an adequate pre-trial investigation.
(ii) Dr. Berry
As noted above, the defense’s theory was that Pavel was not guilty of the charged crimes, and that Matthew and David were accusing Pavel of sexual abuse because they had been manipulated into doing so by their mother, Ms. Pavel, who wanted to gain custody of them. See ante n. 3.
If the factfinder — here, the state trial judge — were to accept this defense, he would have to accept as true an extraordinarily unflattering portrait of Ms. Pavel. But the only evidence presented at trial that supported such a view of Ms. Pavel was Pavel’s general speculations about her character — speculations that were necessarily self-serving and that would inevitably be discounted as such by the judge. Meltzer’s failure to introduce any evidence from a disinterested source in support of the theory that Ms. Pavel was manipulating the boys was a substantial gap in the defense’s presentation. And, as Meltzer knew before trial, that gap could have been filled easily by calling Dr. Berry.
Dr. Berry was a psychiatrist and court-appointed mediator who conducted a number of counseling sessions in September and October of 1988 with Pavel and Ms. Pavel to help them resolve their custody dispute over the boys. Dr. Berry would have been prepared to testify at trial as follows:
It ... appeared to me that the stress [Ms. Pavel] experienced as a result of the marital situation resulted in problems with [Ms. Pavel’s] psychological functioning and memory. For example, [Ms. Pavel] would take very extreme positions regarding visitation and subsequently deny having taken those positions.
Further, [Ms. Pavel] showed a historical distrust and dislike of men that negatively affected her judgment and sense of reality and resulted in paranoia. When disagreed with, [Ms. Pavel] showed a rapid propensity to accuse [Pavel] and me of colluding against her. [Ms. Pavel] also accused [Pavel] of marital rape although, when pressed, she admitted that she did not express to [Pavel] her disinterest or reluctance to engage in sexual relations with him.
This testimony would have buttressed Meltzer’s argument that Ms. Pavel fabricated her trial testimony by implying that she was hostile to the accused. (Indeed, Dr. Berry’s testimony would have suggested that Ms. Pavel’s hostility to Pavel had previously manifested itself in seemingly unfounded claims that he was a sex abuser.)
It would have been within the discretion of the trial court to admit such evidence of Ms. Pavel’s hostility, see post n. 21, and it should have been obvious to Meltzer that such evidence would have been useful to *223the defense.16 See generally Lindstadt, 239 F.3d at 203 (describing, in a similar context, the “exceeding importan[ce]” in a “credibility contest” of “the testimony of neutral, disinterested witnesses”) (internal quotation marks omitted); Crisp, 743 F.2d at 585 (noting the importance of “[h]aving independent witnesses corroborate a defendant’s story”). Meltzer’s explanation for why he opted not to put Dr. Berry on the stand is inadequate, see ante at II. C.l.b, and on appeal respondents do not advert to any possible strategic consideration that could conceivably have counseled against having Dr. Berry testify. Meltzer’s decision not to call Dr. Berry was therefore another factor that contributes to our conclusion that counsel’s cumulative errors made his performance constitutionally deficient. See, e.g., Hart v. Gomez, 174 F.3d 1067, 1071 (9th Cir.1999) (holding that counsel provided constitutionally ineffective assistance where, inter alia, “[h]aving chosen to pursue [a particular] line of defense,” counsel did not introduce readily-available evidence that would have corroborated that line of defense, and there was no plausible strategic reason for his not introducing the evidence).
c. Failure To Call A Medical Expert
Finally, we believe that Meltzer’s performance was deficient to the extent that he did not call a medical expert to testify as to the significance of the physical evidence presented by the prosecution. This decision might well have been beyond reproach if it had been based on appropriate strategic considerations, or if it had been made by Meltzer following a sufficient investigation. But that was not the case.
First, as noted above, Meltzer decided not to put on a medical expert for reasons that had nothing to do with serving Pavel’s interests. See ante ILC.l.b. Therefore, the considerations that animated Meltzer’s decision not to call a medical expert cannot be described as truly “strategic.”
Second, Meltzer’s decision not to call a medical expert was deficient because it was not based on pre-trial consultation with such an expert. See generally ante n. 13 (describing the relationship between pre-trial investigation and the propriety of decisions related to matters of trial strategy). Meltzer should have consulted an expert; as to why, Lindstadt is again relevant.
In that case, as noted above, see ante n. 15, a defendant was convicted of sexually abusing his daughter. The only physical evidence that the prosecution introduced was the testimony of a physician, who testified (1) that he observed some bumps, clefts, and scars in the alleged victim’s vaginal area, and (2) that a “Boston study” and a “review” conducted by another physician established that there was a link between sexual abuse and the bumps, clefts, and scars. The defense attorney failed to obtain either the “Boston study” or the “review,” and he failed to consult *224with an expert before trial about the prosecution’s physical evidence. We characterized this pair of failures as one of the two “[more] compelling” sets of errors that supported our conclusion that the trial attorney had provided constitutionally ineffective assistance.
This case is substantially similar to Lindstadt. Both cases were essentially “credibility contests”: In both cases, the only witnesses to the alleged abuse were its victims and the defendant, and there was no substantial circumstantial evidence of abuse. When a sex abuse case boils down to such a “credibility contest,” physical evidence will often be important. Indeed, “many” sex abuse cases are “close ... on the evidence,” Swofford v. Dobucki, 137 F.3d 442, 443 (7th Cir.1998), and when a case hinges all-but-entirely on whom to believe, an expert’s interpretation of relevant physical evidence (or the lack of it) is the sort of “neutral, disinterested” testimony that may well tip the scales and sway the fact-finder. Williams, 59 F.3d at 682 (“In a credibility contest, the testimony of neutral, disinterested witnesses is exceedingly important.”). Because of the importance of physical evidence in “credibility contest” sex abuse cases, in such cases physical evidence should be a focal point of defense counsel’s pre-trial investigation and analysis of the matter. And because of the “vagaries of abuse indicia,” such pre-trial investigation and analysis will generally require some consultation with an expert.17
For two reasons, that requirement applied here. First, there is no indication in the record that Meltzer had the education or experience necessary to assess relevant physical evidence, and to make for himself a reasonable, informed determination as to whether an expert should be consulted or called to the stand. Cf. United States v. Tucker, 716 F.2d 576, 581 (9th Cir.1983) (holding that, in a complex fraud case, “it should have been obvious to a competent lawyer that the assistance of an accountant [was] necessary”); Knott, 671 F.2d at 1212-13 (noting that counsel may be found to be ineffective for failing to consult an expert where “there is substantial contradiction in a given area of expertise,” or where counsel is not sufficiently “versed in a technical subject matter ... to conduct effective cross-examination”).
Second, there is an obvious, commonsense mismatch — which Meltzer himself recognized before trial — between (1) Matthew and David’s medical indications (nothing abnormal about Matthew, and mild “redness” in David’s anal area) and (2) the allegations against Pavel (that he anally sodomized the boys once a week for four months, and that he did so with even greater frequency in the days leading up to the examination of the boys).18 But in the face of this glaring mismatch a reasonably professional attorney would not have sat on his hands, confident that his client would be acquitted. He would have consulted and been prepared to call an expert *225to drive this disparity home. See generally Holladay v. Haley, 209 F.3d 1243, 1251-52 (11th Cir.2000) (holding that an attorney’s investigation is not “reasonable” within the meaning of Strickland when the facts of a case supply him with “notice” that a particular line of pre-trial investigation may substantially benefit his client, and he does not pursue it).
As to this point, the Eleventh Circuit’s decision in Holsomback v. White, 133 F.3d 1382 (11th Cir.1998), is instructive. In that case, the habeas petitioner was convicted in state court of crimes related to the sex abuse of his children. The petitioner claimed that he received ineffective assistance because “his trial counsel! ] fail[ed] to conduct any [pre-trial] investigation into the conceded lack of medical evidence [against the petitioner], including [trial counsel’s] failure to consult with any physicians concerning the significance of the lack of medical evidence in the case.” See id. at 1386. The Eleventh Circuit agreed, see id. at 1387, explaining:
Although [the petitioner] had been accused of sodomizing his son repeatedly over a period of several years beginning when [the boy] was only four years old and continuing until he was nine, a rectal examination performed shortly after the last alleged incident of sodomy found no medical evidence of sexual abuse. Despite this apparent inconsistency, however, counsel consulted no physician in order to ascertain the significance of the lack of medical evidence. Moreover, ... despite the fact that counsel himself viewed the case essentially as a swearing match, turning entirely on whether the “jury believefd] the child,” he made no effort to support [the petitioner’s] claim of innocence with disinterested medical testimony or other medical evidence suggesting that [the boy’s]'allegations were not credible. In particular, counsel declined to contact either the physician who had examined [the boy] shortly after the last alleged incident of sodomy or [the boy’s] family physician during the period of the alleged sexual abuse, both of whom were readily identifiable to him as potential sources of such disinterested testimony....
... Had counsel interviewed the doctors, a subsequent tactical decision not to call them might have fallen well within “the wide range of reasonable professional assistance.” Strickland, 466 U.S. at 689, 104 S.Ct. 2052. Having conducted no investigation into the significance of the lack of medical evidence that [the boy] had been sexually abused, however, [the petitioner’s] counsel could not have made an informed tactical decision that the risk that the doctors might equivocate on the stand outweighed “what potential benefit might come from [their testimony].” Because counsel never actually spoke with the physicians, he remained entirely unaware ... of whether and to what extent their testimony might have helped [the petitioner’s] case. In these circumstances, we cannot say that counsel’s decision not even to contact the physicians as part of his pretrial investigation was professionally reasonable.
Id. at 1387-88; cf. Schell v. Witek, 218 F.3d 1017, 1028-30 (9th Cir.2000) (in a case in which fingerprint evidence was the “only” evidence against the defendant, remanding for an evidentiary hearing as to whether defense attorney failed to consult with a fingerprint expert before trial, and stating that such a failure could have constituted ineffective assistance of counsel).
In. light of the cumulative weight of the three serious flaws in Meltzer’s representation of Pavel described above, we hold *226that Meltzer’s representation did not fall within the Sixth Amendment’s “wide range” of adequate assistance. See Strideland, 466 U.S. at 690, 104 S.Ct. 2052. We must therefore consider whether Pavel was prejudiced by Meltzer’s poor lawyer-ing.
2. Prejudice
Under Strickland, an ineffective assistance of counsel claimant “must show that there is a reasonable probability that, but for counsel’s unprofessional errors, the result of the proceeding would have been different. A reasonable probability is a probability sufficient to undermine confidence in the outcome.” Id. at 694, 104 S.Ct. 2052.
Here, the standard is met. As an initial matter, we note that in attempting here to show that Meltzer’s performance was not deficient, the State emphasizes that the evidence that it mustered at trial against Pavel was relatively weak. See, e.g., ante n. 18. We agree. It therefore follows that confidence in the judgment of conviction entered against Pavel may be “undermine[d]” here by a relatively smaller quantity of prejudice than might ordinarily suffice.
With this point in mind, we note that each of the above-described flaws in Melt-zer’s representation, see ante II.C.l, prejudiced Pavel.
First, if Meltzer had so much as attempted to prepare a defense here, one of his initial steps would presumably have been to find ways to poke holes in the testimony of Ms. McKinstry. He would have been able to do so. Attached to Pavel’s habeas petition is an affidavit from Dr. Sandra Kaplan, a physician at Cornell University Medical College. Dr. Kaplan’s affidavit states that Ms. McKinstry’s evaluation of the boys was conducted in a manner that was flatly inconsistent with the relevant, publicly available guidelines of the American Academy of Child and Adolescent Psychiatry (“the Guidelines”). See Lindstadt, 239 F.3d at 201-02 (assuming that had the relevant trial attorney performed an adequate pre-trial investigation, the defense would have discovered various articles published in scientific journals). For example, Dr. Kaplan noted that Ms. McKinstry was unqualified to evaluate allegations that the boys had been abused because she (1) lacked requisite professional qualifications, and (2) was serving as the boys’ therapist, and “[a]s the boys’ therapist, Ms. McKinstry had a professional obligation not to serve as as an evaluator with respect to allegations of sexual abuse involving her patients.” Moreover, Dr. Kaplan noted that in cases of alleged intra-family sex abuse such as this one, in which there are discernible “indicia of false allegations,”19 the Guidelines require an “evaluator” of abuse allegations to “obtain a history from the perspective of each parent” (which Ms. McKinstry did not do) and/or to conduct joint sessions with both parents and the children present (which, again, Ms. McKinstry did not do). Similarly, Dr. Kaplan noted that the boys were interviewed repeatedly concerning their father’s alleged abuse, and that “the [Gjuide-lines warn [that] multiple interviews merely encourage the child to create a story to meet the demands of [interviewing] adults for more information.”20 See generally, *227e.g., Jacqueline Miller Beckett, Note, The True Value of the Confrontation Clause: A Study of Child Sex Abuse Trials, 82 Geo. L.J. 1605, 1606 (1994) (“many children have trouble distinguishing fact from fiction and are willing to lie if they are rewarded for telling such stories”) (collecting sources, id. at 1633-37).
Second, for the reasons stated above, see ante II.C.l.b.(i)-(ii), if Meltzer had put Clothilde Pavel and Dr. Berry on the stand, their testimony would have substantially buttressed the defense’s theory of the case — even if a skilled prosecutor could have suggested that some of the details of Clothilde Pavel’s testimony were presented in an arguably hyperbolic fashion.21 See Affidavit of Clothilde Pavel ¶ 5 (executed March 27, 1992) (stating that Clothilde Pavel* Pavel, and the boys “literally spent all of our time together” during the Florida trip) (emphasis added).
Third, had Meltzer conducted an adequate pre-trial investigation of the physical evidence against Pavel, and sought a medical expert who might have testified on behalf of the defense, there can be little doubt that Meltzer would have come across and called to the stand a physician such as Dr. Margaret McHugh- — a physician with extensive experience in, inter alia, evaluating and/or treating victims of child abuse.22 In an affidavit submitted in the state-court post-conviction proceedings, Dr. McHugh stated unequivocally that the boys’ medical condition was simply not “consistent” — as Dr. Madden had *228testified — with their having been repeatedly anally sodomized with the frequency that they described:
I have reviewed the Grand Jury and trial testimony of Matthew and David Pavel....
It is my opinion, to a reasonable degree of medical certainty, that the clinical findings regarding Matthew and David are inconsistent with their testimony. Repetitive sexual attacks of the nature described by the boys, by an adult male without the use of a lubricant and/or sedation would be expected to result in clinical evidence such as tearing injuries and abrasions to each child’s anus and rectum which would [have been] apparent on gross examination.
I have also reviewed the testimony of Dr. Celeste Madden. Dr. Madden testified that the absence of any abnormal findings in Matthew’s physical examination is not inconsistent with the crimes of sodomy and sexual abuse. Given the boys’ testimony, however, it is highly improbable that such repeated attacks within five (5) days of the exam would leave no physical evidence.
... [As to David, on whom a “mild perianal erythema” was noticed,] “[c flearly, the numerous acts of sodomy described by David would be expected to result in more than a mild erythema, i.e., a slight red discoloration.
(Emphasis Added)
In light of the weakness of the prosecution’s trial evidence, see ante at 225, Dr. McHugh’s opinion casts a long shadow over this case.23 And in combination with the other points described above, see ante at 225-28, it leads us to a disturbing conclusion:. Had Meltzer performed in a constitutionally effective manner, there is — at the very least — a “reasonable probability” that Pavel would not have been convicted of the crimes with which he was charged, and for which he has been punished these past eleven and a half years.
Before concluding, we pause briefly to commend Kelley Drye & Warren, which took on this matter pro bono publico over eight years ago, and has now seen it through. Kelley Drye’s work on this case has been tenacious and consistently skillful, and by providing their client with superbly effective assistance of counsel, the firm has acted in the best traditions of our profession.
IV. Conclusion
For the reasons stated above, we hold that the state judgment of conviction *229against petitioner-appellant Kenneth G. Pavel was entered in violation of his rights under the Sixth Amendment to the United States Constitution. Accordingly, we
(1) reverse the judgment of the District Court,
(2) vacate the state court judgment of conviction,
(3) and remand the cause.
On remand, the District Court shall issue a writ of habeas corpus to Mr. Pavel on the thirtieth calendar day after the issuance of our mandate unless New York State has, by that point, taken concrete and substantial steps expeditiously to retry Pavel.
The mandate shall issue forthwith. If further proceedings arising from Pavel’s habeas petition are required in this Court, the parties shall inform the Clerk of this Court. Jurisdiction will then be automatically restored to this Court without need for a new notice of appeal. See United States v. Jacobson, 15 F.3d 19, 21-22 (2d Cir.1994). After jurisdiction is restored, the Clerk shall set an expedited briefing schedule, and the matter will then be heard by this panel on letter briefs.