HARLINGTON WOOD, Jr., Circuit Judge.

The state appeals the district court’s grant of a writ of habeas corpus for Larry Cosey, who was convicted in an Illinois state court of attempted murder, armed robbery, and aggravated battery. The district court held that Cosey was denied his constitutional right to effective assistance of counsel. This case was before this court once before, United States ex rel. Cosey v. Wolff, 682 F.2d 691 (7th Cir.1982) (per curiam), where we reversed and remanded an earlier district court grant of summary judgment to Cosey in his petition for a writ of habeas corpus. We held that factual issues and fairness demanded that both sides be heard. Id. at 693. The district court has held such a hearing, and we now affirm the order granting the writ.

The facts of this case have been recounted in much detail in a number of published opinions. 1 We repeat only the background information necessary for an understanding of this appeal. The prosecution’s case against Cosey and co-defendant Ernest Van Johnson relied primarily on the victim’s testimony. The victim identified Cosey and his co-defendant as the assailants. The defense counsel, Joseph Stillo, attempted to discredit the victim’s testimony by pointing out inconsistencies in his story, and by noting that he was a drug addict. After the judge denied the defense’s motion for acquittal, Stillo had Cosey and his co-defendant take the stand to deny their guilt. Stil-lo -called no one else to testify on behalf of the defendants.

Sometime before trial Cosey told Stillo that he had five persons who would testify in his favor. Cosey’s stepfather and a tenant who lived in the building where the shooting allegedly took place would have testified that they were home on the evening in question and heard no unusual noises. Cosey’s mother and a carpenter would have testified that the basement, the alleged scene of the crime, was being remodeled at the time, and they noticed no change in its condition. Finally, Johnson’s girlfriend was ready to testify that she and Johnson were together on the evening the crime took place. Stillo failed to so much *658 as interview or investígate even one of these potential witnesses. Stillo’s failure allegedly resulted from his professional judgment that the state’s case against Co-sey was so weak that it would be entirely unnecessary to call any witnesses to the stand other than Cosey and his co-defendant. The state judge apparently did not share Stillo’s assessment of the state’s case, and found Cosey guilty on all counts after a bench trial. 2

The state now attempts to argue that Stillo’s out-of-hand rejection of the proffered witnesses without even interviewing or investigating them did not fall below minimum standards of professional competence. While we agree that in particular cases an attorney’s professional judgment may lead him to decline to interview persons identified by his client as potential witnesses, we agree with Judge Getzendanner that this was not such a case for the reasons that she states. See United States ex rel. Cosey v. Wolff, 562 F.Supp. 140 (N.D.Ill.1983). 3

We must now address the issue of prejudice. The government notes that the Illinois Court of Appeals did not believe that the absence of the witnesses at trial prejudiced the defendants “since the trial court would not necessarily have rendered a different verdict.” People v. Cosey, 82 Ill. App.3d 968, 38 Ill.Dec. 425, 403 N.E.2d 656, 661 (1980). This is not the correct standard to apply when evaluating whether ineffective assistance of counsel has prejudiced a defendant. Once a court finds that a trial counsel’s performance was constitutionally inadequate, and that the counsel’s acts or omissions may have impaired the defense, the burden of proof is on the government to prove beyond a reasonable doubt that the trial counsel’s ineffective assistance did not affect the verdict. See United States v. Tucker, 716 F.2d 576, 588 (9th Cir.1983) (opinion of Judge Alarcon) (as corrected); Wade v. Franzen, 678 F.2d 56, 59 (7th Cir.1982); United States v. Baynes, 687 F.2d 659, 673 (3d Cir.1982); Wood v. Zahradnick, 578 F.2d 980, 982 (4th Cir.1978); United States v. Decoster, 624 F.2d 196, 208 & n. 74 (D.C.Cir.1976) (en banc) (plurality opinion). 4 This the state failed to do.

*659 The state maintains that Cosey suffered no prejudice because he ultimately got the allegedly exculpatory material before the trier of fact through a petition for a new trial. 5 Cosey had supported his petition with affidavits from the five proffered witnesses setting out what they would have testified to. The state concludes from the trial judge’s rejection of the petition that it is clear beyond a reasonable doubt that Cosey would have been convicted even if the witnesses had presented their testimony during the defendants’ case in chief. We are not persuaded.

The state judge considered the witnesses’ testimony within the context of a motion for a new trial based on newly discovered evidence. The standards for granting a new trial based on newly discovered evidence and based on attorney incompetence are significantly different. Compare United States v. Oliver, 683 F.2d 224, 228 (7th Cir.1982) (newly discovered evidence) with Wade, 678 F.2d at 59 (attorney incompetence); see also Jackson v. United States, 371 F.2d 960, 962 (D.C.Cir.1966). Where a defendant claims a right to a new trial based on newly discovered evidence, he bears the burden of proving, among other things, that the evidence would probably lead to an acquittal. Where a defendant suffered from ineffective assistance of counsel, the government bears the burden of proving beyond a reasonable doubt that competent counsel would not have been able to secure an acquittal.

The trial judge’s rejection of the witnesses’ testimony as grounds for a new trial based on a plea of newly discovered evidence does not convince us beyond a reasonable doubt that Cosey would have been found guilty had his original trial counsel performed competently. The state admitted in oral argument that the witnesses’ testimony as set out in the affidavits, if believed, would have exculpated Cosey. The state asserted that the trial judge evaluated the credibility of the witnesses based on their affidavits, and then rejected them. Nothing in the record, however, indicates why the trial judge denied the motion for a new trial. 6 Moreover, the trial judge held no hearing at which the proffered witnesses were present and cross-examined, and therefore had no reason to disbelieve their sworn statements. 7

The district court’s order granting Co-sey’s petition for a writ of habeas corpus is affirmed.

Notes

. See United States ex rel. Cosey v. Wolff, 682 F.2d 691 (7th Cir.1982) (per curiam); United States ex rel. Cosey v. Wolff, 562 F.Supp. 140 (N.D.Ill.1983); United States ex rel. Cosey v. Wolff, 526 F.Supp. 788 (N.D.Ill.1981); People v. Cosey, 82 Ill.App.3d 968, 38 Ill.Dec. 425, 403 N.E.2d 656 (1980).

. The judge sentenced Cosey to 20 years.

. Although the state’s argument is not entirely clear, the state implies that the district court’s emphasis on Stillo’s failure to interview and investigate the proffered witnesses is misplaced. The state maintains that Stillo did not need to interview the witnesses because Cosey told him what the witnesses would have testified to. The state then implies from Stillo’s knowledge that his decision not to follow through with any of the witnesses must have been the result of some strategy. Appellant’s Brief at 18. Stillo’s knowledge of what the witnesses would have said at trial does not convince us that his refusal to interview them resulted from a professionally competent tactical decision that may simply appear wrong in retrospect; rather, it highlights the incompetence of his failure to seek them out. Cosey’s entire defense at trial rested on discrediting the state’s main witness — the victim. The five proffered witnesses would not only have corroborated Cosey’s story and further impeached the victim’s version, but, as the state conceded in oral argument, if the witnesses were believed, their testimony alone would have entirely exculpated Cosey. Without interviewing and investigating such promising witnesses, Cosey’s attorney had no reason to believe they would not be valuable in securing Cosey’s release. Although three of the witnesses had an apparent reason to be biased in Cosey’s favor, that alone is insufficient cause to automatically reject them. Moreover, two of the proffered witnesses had no apparent reason for bias. There was no strategy involved here, only negligence. See United States v. Tucker, 716 F.2d 576, 593-94 (9th Cir.1983) (opinion of Judge Alarcon) (as corrected).

. The state’s brief and oral argument evidenced some confusion on the defendant’s and the state’s burden of proof with respect to the prejudice issue. The source of this confusion is the state’s erroneous characterization of this case as one in which “the petitioner seeks to obtain habeas corpus relief by only a showing of ‘new evidence’ which defense counsel did not produce at trial.” Classifying the testimony of the five proffered witnesses as “new evidence,” the state quotes United States v. Frady, 456 U.S. 152, 102 S.Ct. 1584, 71 L.Ed.2d 816 (1983), which held that a claim of error concerning new evidence does not rise to constitutional magnitude unless the defendant proves that the error created not only “a possibility of prejudice, but that [it] worked to his actual and substantial disadvantage.” Id. at 170, 102 S.Ct. at 1596 (emphasis in original). This is not, however, a case of new evidence *659 surfacing after trial; it is one of ineffective assistance of counsel before and during trial for failure to interview and investigate these known witnesses. Thus, an error of constitutional magnitude had already been proven. The burden therefore falls on the state to prove beyond a reasonable doubt that despite the error, Cosey would have been convicted.

. The same judge who found Cosey guilty also considered his petition for a new trial.

. The trial judge may, for example, have determined that the witnesses’ testimony did not qualify as “newly discovered evidence” because trial counsel knew of the witnesses’ existence before trial.

. As we explain in note 3, supra, that three of the five witnesses could arguably be biased in Cosey’s favor is insufficient to automatically dismiss all five sworn statements.