The government appeals the issuance of a provisional writ of habeas corpus pursuant to 28 U.S.C. § 2254 for Petitioner Brian Keith Fargo, based on his claim of ineffective assistance of counsel at trial. We reverse.

I.

After a jury trial, Fargo was convicted of third degree criminal sexual conduct, in violation of M.C.L. § 750.520d. The district court held that the state courts that considered Fargo’s post-conviction claim of ineffective assistance of counsel based their decisions on an unreasonable determination of fact and an unreasonable application of clearly established federal law.

Fargo was found guilty of the sexual penetration of Sheila Ruggero, a 15-year old, in Ludington, Michigan on April 23, 1992, sometime between 9:30 and 10:00 p.m. Fargo did not testify at trial, and the defense presented no witnesses. Henry Dongvillo, Fargo’s appointed counsel, argued that Fargo had always maintained his innocence and told the police the same, there was no medical evidence showing that Ruggero had been sexually penetrated, and, accordingly, the jury should apply the facts to the law and find Fargo not guilty. After Fargo was found guilty, he pled guilty to being a fourth felony offender and was sentenced to 15-30 years in prison.

Fargo’s appellate counsel filed a motion for a new trial, requesting a remand for an *605evidentiary hearing on his claim that he received ineffective assistance of counsel. Fargo contended before the trial court during the ensuing evidentiary hearing that Dongvillo had rendered ineffective assistance by failing to adequately investigate, prepare and present an alibi defense. Fargo testified at the hearing that he was not with Ruggero at the time of the alleged assault. He testified that he saw Mary Quinn and Gina Johnson at the beach that evening, around 7:00 p.m., and invited them back to his apartment for drinks. He claimed that the three of them hung out at his apartment until about 9:50 p.m., when they all left together. Fargo testified that he went directly to Dan Wilby’s house, where he stayed until 11:05 or 11:10 p.m. Fargo testified that he told Dongvillo of his alibi for the night in question and gave him the witnesses’ names and contact information.

Dongvillo testified at the evidentiary hearing that Fargo had told him about some alibi witnesses and had furnished him with a list of at least three names and telephone numbers. Dongvillo admitted that he did not contact every person on the list, but recalled speaking with a male and a female at the same number, presumably Wilby and Quinn, who did not have a phone but could be reached through Wilby. Dongvillo testified that he felt that the witnesses would not be helpful because they did not have independent recollections of the dates involved. He testified that he told Fargo they were not good alibi witnesses and did nothing further regarding the alleged alibi defense Dongvillo also testified that he remembered hearing from the prosecutor in the case that Fargo was suggesting to people that they provide untruthful testimony in his favor.

At the end of the evidentiary hearing, the state trial court denied Fargo’s motion for a new trial. The Michigan Court of Appeals affirmed the trial court, and the Michigan Supreme Court denied review.

Fargo filed a petition for habeas corpus. The district court granted Fargo a provisional writ, finding the state court’s decision was an unreasonable factual determination that the witnesses did not provide Fargo with an alibi and an unreasonable legal determination that Dongvillo provided effective assistance in that he admittedly failed to call the third potential alibi witness.

II.

In determining whether to issue a habeas writ, the district court’s review of a state court decision is governed by the standards set forth in the Antiterrorism and Effective Death Penalty Act of 1996, 28 U.S.C. § 2241 et seq. (“AEDPA”). Id. The AEDPA only provides habeas relief for a state prisoner in certain limited circumstances, where the adjudication: “(1) resulted in a decision that was contrary to, or involved an unreasonable application of, clearly established Federal law, as determined by the Supreme Court of the United States; or (2) resulted in a decision that was based on an unreasonable determination of the facts in light of the evidence presented in the State court proceeding.” 28 U.S.C. § 2254(d).

III.

This is an AEDPA case. In this case, AEDPA mandates not just the standard that we apply, but also the result that we reach. AEDPA requires that we accord strong deference to state courts’ application of federal law.

In this case, the government has conceded that the trial court made an unreasonable factual determination in assessing the potential alibi provided by the witnesses that Fargo proffered in his post-*606conviction hearing. The trial court assessed the time line of the proffered witnesses accounting for Fargo’s whereabouts on the night in question based on the unreasonable determination that nightfall occurred “about 7 or 8 p.m.” Nature, as recorded by the Naval Observatory, is not in accord with that assumption. Nightfall did not occur till after 9 p.m. The trial court’s entire assessment of the potential helpfulness of the alibi witnesses depended upon that assumption. A careful examination of the trial court’s holding, however, reveals that recognizing the trial court’s error does not end our analysis.

The claim before the trial court, as here, was the ineffective assistance of Fargo’s trial counsel in not presenting these alibi witnesses. The trial court did not find that based on the above factual determinations that no prejudice inured to Fargo, and, therefore, counsel did not provide ineffective assistance. The trial court held as follows:

The Court puts weight on Attorney Dongvillo’s statement that when he did call the witnesses that he did reach, that their sharing of information would not have provided an alibi defense and certainly that recollection, even though it’s no longer specific as to what they said, would be consistent with the ... scenario that the Court has just recited based upon what two of these witnesses are even saying here today.

The court does not find that Attorney Dongvillo failed to properly investigate and the Court does not find that the, so to speak, new witnesses are in fact new witnesses. The Court [denies] the Motion for New Trial.

Thus, in determining that counsel conducted a reasonable investigation of the potential alibi defense, the trial court actually credited counsel’s testimony that he contacted two of the three witnesses and determined that they did not provide a viable alibi defense for Fargo. In his testimony, counsel did not state for the record the reasoning that the trial court suggested for his decision, ie., the scenario regarding the time differential in the alibi’s accounts that would have created a window of opportunity for the crime to be committed. Rather, counsel stated that the witnesses could provide no specifies. After speaking with these witnesses, he concluded that neither of them had an independent recollection of the times and places that they had been with Fargo: the persons he spoke to “didn’t know the dates that we’re talking about and they weren’t even sure about the person that we were talking about.” Counsel also testified that, after interviewing these two potential witnesses, he believed that the only way they would be useful alibi witnesses was “if somebody gave them the detailed information that they needed to testify to and I wasn’t about to do that.” His conclusion perhaps also took into account the warning he received from the prosecutor about rumors that Fargo was attempting to produce a false alibi. Accordingly, the trial court’s dubious discussion of the “scenario” under which Fargo could have committed the crime despite the witnesses’ account of some of his activities that night constituted no more than dicta in support of counsel’s determination that such witnesses would not have provided an alibi. The question before the district court and on appeal is whether the trial court’s finding that counsel conducted a reasonable investigation of the alibi witnesses was unreasonable.

While the state court did not mention the case by name, the state court’s opinion can be read as having applied the clearly established federal law for deciding ineffective assistance of counsel claims found in Strickland v. Washington, 466 U.S. 668, 104 S.Ct. 2052, 80 L.Ed.2d 674 (1984). *607Under Strickland, a petitioner must show both (1) performance of counsel that was so deficient it falls “below an objective standard of reasonableness” and (2) resulting prejudice to his case. Id. at 686-87, 104 S.Ct. 2052. The district court read the state court decisions in the present case as holding that Fargo had failed to establish that Dongvillo’s performance was deficient. On collateral review, then, the question is whether it was unreasonable to find that counsel acted satisfactorily in his investigation of the alibi defense. There are two components to consider in evaluating this investigation: (1) the investigation that he did conduct in talking to Wilby and Quinn and finding their testimony of no value, and (2) the investigation that he did not conduct in failing to contact the third witness, Johnson.

The district court held that it was unreasonable for the trial court to have credited counsel’s assertion that he had actually contacted Wilby and Quinn. The district court noted that all three alibi witnesses at the evidentiary hearing claimed that they were not contacted, and counsel had no specific information to back up his belief that he had called two of them. But, federal courts must give state court factual determinations a presumption of correctness that can be overcome only by clear and convincing evidence to the contrary. 28 U.S.C. § 2254(e)(1) (“[A] determination of a factual issue made by a state court shall be presumed to be correct. The [habeas] applicant shall have the burden of rebutting the presumption of correctness by clear and convincing evidence.”); see also West v. Seabold, 78 F.3d 81, 83 (6th Cir. 1996). Both the district court in its decision and Fargo in this appeal stress that counsel could not produce his file at the hearing reflecting any notes on his conversations with these witnesses or the list of witnesses submitted to him by Fargo. The state court of appeals noted, however, that counsel had no specific information at his disposal because Fargo’s appellate counsel had failed to compel him to bring his files. The trial court credited counsel’s testimony that he contacted them. The state court’s determination that counsel’s testimony was more credible than the witnesses’ testimony was not unreasonable. There is no clear and convincing evidence to justify a rejection of the trial court’s credibility finding.

The second component of the investigation is troublesome. Counsel candidly admitted in his testimony before the trial court that Fargo submitted a list of at least three purported alibi witnesses, and he contacted only two of those witnesses. Were we reviewing this case on direct appeal, we might very well find that the failure to contact a clearly identified alleged alibi witness including contact information (Fargo provided Johnson’s phone number) constitutes a deficient, unreasonable investigation by counsel. As we alluded to in the beginning of our analysis, however, AEDPA governs the result here. Under the facts before us, we cannot say that the state courts were unreasonable in finding that this failure did not render the investigation unreasonable. We know from the evidentiary hearing testimony that Johnson and Quinn gave the same story as to their interaction with Fargo on the date of the crime, and it is not unreasonable to conclude that counsel could have determined that after contacting Quinn and finding her account dubious and unhelpful that contacting Johnson would be unnecessary.

As the dissent suggests, the state court erroneously determined the time of nightfall, and concluded that there was a window of time for the criminal act between 7:00-9:30 P.M. or 8:00 to 10:00 P.M. Nevertheless, there is still a window of time in *608which Fargo could have committed the rape. If Quinn and Johnson left Fargo’s apartment before dark, as Johnson testified, that would be around 9:00 P.M. If Fargo arrived at Wilby’s apartment between 9:30-10:00 P.M., stayed about one or one-and-a-half hours (according to Wilby), and left around 11:05 or 11:10 P.M. (according to Fargo), Fargo would still have been available to commit the act in his own apartment between around 9:00-10:00 P.M. Thus, although the state court decided the time window was after the visit to Wilby’s apartment, it was likely between the time when Quinn and Johnson left Fargo’s apartment and when Fargo went to the Wilby apartment.

The record at the habeas corpus hearing does suggest that [counsel] could well have made a more thorough investigation than he did. Nevertheless, in considering claims of ineffective assistance of counsel, we address not what is prudent or appropriate, but only what is constitutionally compelled. We have decided that strategic choices made after less than complete investigation are reasonable precisely to the extent that reasonable professional judgments support the limitations on investigation. Applying this standard, we agree with the courts below that counsel’s decision not to mount an all-out investigation ... was supported by reasonable professional judgment.

Burger v. Kemp, 483 U.S. 776, 794, 107 S.Ct. 3114, 97 L.Ed.2d 638 (1987) (citations and quotations omitted). We cannot find the state courts’ legal determination unreasonable in light of the facts.

REVERSED.

dissenting.

The majority here concludes that the Michigan state courts were not unreasonable in their decision that Fargo was not denied effective assistance of counsel. Because I believe the majority to be in error, I respectfully dissent.

At the evidentiary hearing in state court, Fargo presented three witnesses, none of whom testified at trial, who would have testified that they were with Fargo the night of the assault - thus giving Fargo a valid alibi. Despite the fact that Fargo gave his attorney, Henry Dongvillo, their contact information, each of them testified that Dongvillo never contacted them.

The majority supports its conclusion by noting the presumed correctness of the state court’s belief that Dongvillo was correct when he claimed that he contacted two of those witnesses, Quinn and Wilby, and said that they had little to offer in Fargo’s defense. But even assuming the correctness of this statement (which seems on its face questionable), no one disputes that Dongvillo candidly admitted that he never contacted the third alibi witness, Gina Johnson. The majority argues that Dongvillo’s failure could make some sense given that “[w]e know from the evidentiary hearing testimony that Johnson and Quinn gave the same story as to their interaction with Fargo on the date of the crime, and it is not unreasonable to conclude that counsel could have determined that after contacting Quinn and finding her account dubious and unhelpful that contacting Johnson would be unnecessary.” Maj. op. at 7.

This cannot excuse Dongvillo’s behavior. The fact that Dongvillo found Quinn’s account dubious and unhelpful does not in any way suggest that he would have found the same for Johnson’s testimony. Dongvillo believed that Quinn was an un*609helpful witness because she had no detailed information of what happened that night. Dongvillo had no reason to suspect that Johnson would be plagued with similar memory lapses. In truth, the fact that Dongvillo found Quinn an ineffective witness should have only intensified (rather than diminished) the need to contact Johnson. And the fact that Quinn and Johnson ultimately gave similar testimony in a later proceeding does not mitigate Dongvillo’s substandard performance. It is plain fact that if Dongvillo had contacted Johnson, he would have discovered a presumably credible witness (rather than Quinn, whom he believed not to be credible) who would have testified to Fargo’s alibi. There is simply no explanation for Dongvillo’s failure to contact a witness presented to him as one that could completely exculpate his client.

As regards prejudice, the state courts believed there was no prejudice because the alibi witnesses, even if believed, could not have conclusively established an alibi defense. Wilby did not meet up with Fargo until 9:80 to 10:00 p.m. Quinn and Johnson, according to Johnson, left Fargo slightly after dark. Interpreting “dark” as being between 7:00 and 8:00 p.m., the state court reasoned that this left at least an hour and a half for the sexual assault to occur. We know now that these calculations are indisputably erroneous. For sunset, according to the United States Naval Observatory, did not occur on that day until 8:40 p.m. Civil twilight - a time at which earthly objects can still be clearly distinguished, according to the Naval Observatory - did not occur until 9:11 p.m. Complete darkness did not follow until significantly afterward. The majority argues that there is still a window of time in which Fargo could have committed the assault consonant with the testimony of his alibi witnesses. Quinn and Johnson did not leave Fargo at his place until after 9:11 p.m., according to Johnson. (Quinn actually went further, testifying that they did not leave before 9:30 p.m.) Fargo met up with Wilby at Wilby’s place between 9:30 and 10:00 p.m. This leaves, at best, a nineteen to forty-nine minute window in which Fargo could operate. The majority contends that in that window, Fargo would have time to leave his house, pick up Ruggero from her house, take her back to his place, watch television with her on the couch, “make out” (in her words) with her consensually on the couch, then turn on Ruggero, drag her back to his bedroom, sexually assault her, get dressed, drive Ruggero to the home of her friend Adeline Warner, and then get to Wilby’s house. This seems grossly implausible. And even if Fargo could somehow accomplish those tasks in the time allotted, the testimony of Fargo’s alibi witnesses unavoidably conflicts with the prosecution’s account of the facts in major ways. For example, prosecution witnesses testified that Fargo picked up Ruggero between 8:30 and 9:00 p.m. and did not return her until 10:30 p.m. Fargo’s alibi witnesses flatly contradict those times. Of course, I do not mean to insinuate that prosecution’s account of the events is necessarily the wrong one - I only mean to point out that the testimony of the witnesses that Dongvillo should have called would have been incompatible with such an account. The jury would still have to believe the testimony of these alibi witnesses over the prosecution’s witnesses, of course, but the failure to call them at all certainly creates “a reasonable probability that, but for counsel’s unprofessional errors, the result of the proceeding would have been different.” Strickland v. Washington, 466 U.S. 668, 694, 104 S.Ct. 2052, 80 L.Ed.2d 674 (1984). As a result, Fargo has made out such a case of both deficiency and prejudice under Strickland that the Michigan courts’ decision holding other*610wise is objectively unreasonable. I respectfully dissent.