*462 Opinion of the Court
COX, Chief Judge:Appellant was tried by a general court-martial composed of officer members at Fort Ord, California, on October 15 and 17-19, 1990. Contrary to his pleas, he was convicted of committing indecent acts with a child under the age of 16 (3 specifications), in violation of Article 134, Uniform Code of Military Justice, 10 USC § 934. He was sentenced to a dishonorable discharge, confinement for 14 years, total forfeitures, and reduction to E-l. The convening authority approved the adjudged sentence. The Court of Criminal Appeals affirmed the findings and sentence in an unpublished opinion.
We granted four issues for review.1 We resolve this case on the first granted issue and reverse the decision of the Court of Criminal Appeals.
Discussion
This case has a long appellate history. Six years have passed since appellant’s original trial. This lengthy period of time has been consumed by several trips back and forth between the appellate courts. In our initial decision on August 3,1993, we remanded the case for additional factfinding in order to facilitate our consideration of appellant’s claim of ineffective assistance of counsel. United States v. Wean, 37 MJ 286 (CMA 1993). On June 7, 1994, the then-Court of Military Review2 ordered the original trial defense counsels to answer four specific questions posed by this Court in our remand opinion.3 On August 3, 1994, after realizing that trial defense counsels’ answers, combined with the record of trial, did not resolve the issues, the Court of Military Review ordered a limited hearing in lieu of additional affidavits pursuant to United States v. DuBay, 17 USCMA 147, 37 CMR 411 (1967). See Appendix A.
Specific findings of fact and conclusions of law were made by the military judge who presided over the DuBay hearing. See Appendix B. After considering all of the evidence adduced at the DuBay hearing, as well as the entire record of trial, the military *463judge concluded that “the defense representation of appellant was clearly deficient,” that this deficient representation “involved the essential issues in the case,” and that it had a “clear prejudicial impact” on appellant’s right to a fair trial. Appendix B at 9. In affirming appellant’s convictions in an unpublished opinion, however, the court below rejected most of the military judge’s findings of fact and his entire conclusions of law.
It is well-established that a determination regarding the effectiveness of counsel is a mixed question of law and fact. Strickland v. Washington, 466 U.S. 668, 698, 104 S.Ct. 2052, 2070, 80 L.Ed.2d 674 (1984); S. Childress and M. Davis, 2 Federal Standards of Review § 12.09 at 62-63 (2d ed.1992). Therefore, the factual findings of the military judge are reviewed under a clearly-erroneous standard, and the ultimate determinations whether the representation was ineffective and, if so, whether it was prejudicial are reviewed de novo. Buenoano v. Singletary, 74 F.3d 1078, 1083 (11th Cir. 1996), citing United States v. Camacho, 40 F.3d 349, 353 (11th Cir.1994); see Oliver v. Wainwright, 782 F.2d 1521, 1524 (11th Cir. 1986). The test for ineffective assistance of counsel is as follows:
First, the defendant must show that counsel’s performance was deficient. This requires showing that counsel made errors so serious that counsel was not functioning as the “counsel” guaranteed the defendant by the Sixth Amendment. Second, the defendant must show that the deficient performance prejudiced the defense. This requires showing that counsel’s errors were so serious as to deprive the defendant of a fair trial, a trial whose result is reliable.
Strickland, supra at 687, 104 S.Ct. at 2064; United States v. Ingham, 42 MJ 218, 223 (1995).
The DuBay judge found several facts which he used as the basis for his conclusion that the assistance of counsel in this case was ineffective. See Appendix B. Reviewing these facts under a clearly erroneous standard, we conclude that they are well-supported by the record of trial, and therefore we accept all of these factual findings for purposes of our analysis. Also, we necessarily defer to the DuBay judge’s determinations of credibility in this regard. See United States v. Williams, 37 MJ 352, 357 (CMA 1993); United States v. White, 36 MJ 284, 287 (CMA 1993). Our task here is greatly facilitated by the high quality of the DuBay record.
Therefore, we turn to the de novo portion of our review and decide whether these facts support a finding of deficiency, and if so, whether this deficiency rises to the level of prejudice.
The Court of Criminal Appeals held that the DuBay judge erred in his findings that “the defense representation on behalf of the appellant ... ‘involved essential issues in the case,’ ” and was “clearly deficient.” Unpub. op. at 6-7 n. 3 and 10-11. United States v. Wean, supra at 6-7 n. 3 and 10-11. We disagree with the lower court and concur with the DuBay judge’s finding that counsel’s performance was “clearly deficient” in at least three particulars. First, defense counsel’s approach to the use of expert witnesses by the Government, coupled with his omission in not using expert testimony, demonstrated a lack of understanding of the law and a failure to properly research and investigate appellant’s case. Second, the division of responsibilities between the civilian defense counsel and detailed military counsel was at the “heart of the defense problem in representing the appellant____ [T]here was essentially no discussion between the two regarding the issues, the law, or the case methodology appropriate for the representation of the appellant.” Third, civilian defense counsel inappropriately made a comment during the sentencing argument that appellant was suffering from “an illness of the mind [which] compelled him to do these things,” with absolutely no basis in fact for the comment. This comment was directly contrary to appellant’s insistence upon his innocence.
We thus turn to the second prong of the Strickland test to consider whether ap*464pellant was prejudiced by the deficient performance. Prejudice occurs when the defendant shows 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.” Strickland, supra at 694, 104 S.Ct. at 2068. “Moreover, a verdict or conclusion only weakly supported by the record is more likely to have been affected by errors than one with overwhelming record support.” Id. at 696, 104 S.Ct. at 2068.
Unlike the Court of Criminal Appeals, we find that prejudice is apparent from the Du-Bay record itself. Here, appellant denied guilt all along and continues to do so. The case against him consisted of untested theories of “play therapy” and hearsay statements, which were barely contested at trial by appellant’s counsel. The physical evidence did not conclusively point to appellant. Assuming arguendo that the children were sexually abused, other individuals besides appellant also had access to them. The initial identification of appellant was made by only one child in this ease, and was suggested to the other children by questions such as, “[Did] Steven play[ ] any games where he touched [you]?” In other words, the entire case against appellant consisted of hearsay testimony admitted under various exceptions.
It is certain from this fact alone that if defense counsel had attacked admissibility of the underlying theories of the government experts’ opinions and had interviewed any of appellant’s proposed witnesses, they would have at least altered the landscape of the trial. Moreover, in general, when an accused has consistently denied guilt, a functional defense counsel should not concede an accused’s guilt during sentencing, not only because this can serve to anger the panel members, but also because defense counsel may be able to argue for reconsideration of the findings before announcement of the sentence.4 The tenor of civilian defense counsel’s responses to the DuBay judge’s inquiries was such that it was clear the deficiencies in his performance were not due to tactical decisions but, rather, were due to his failure to investigate, prepare, and research. See United States v. Scott, 24 MJ 186 (CMA 1987). As a matter of law, the result of this trial was unreliable.
The decision of the United States Army Court of Criminal Appeals is reversed. The findings and sentence are set aside, and the record of trial is returned to the Judge Advocate General of the Army, who may return it to an appropriate convening authority to order a rehearing, or if a rehearing is impractical, to dismiss the charges against appellant. Art. 67(e), UCMJ, 10 USC § 867(e) (1994).
Judges CRAWFORD, GIERKE, and EFFRON concur.. The following issues were granted:
I
WHETHER STAFF SERGEANT WEAN WAS DENIED HIS SIXTH AMENDMENT RIGHT TO EFFECTIVE ASSISTANCE OF COUNSEL ON THE MERITS BECAUSE DEFENSE COUNSEL FAILED TO INVESTIGATE APPELLANT’S CASE AND ADEQUATELY PRESENT APPELLANT’S DEFENSES AT TRIAL.
II
WHETHER STAFF SERGEANT WEAN WAS DENIED HIS SIXTH AMENDMENT RIGHT TO EFFECTIVE ASSISTANCE OF COUNSEL ON THE MERITS BECAUSE DEFENSE COUNSEL'S CLOSING ARGUMENT ON SENTENCING INCLUDED IRRATIONAL ASSERTIONS SUBSTANTIALLY PREJUDICIAL TO STAFF SERGEANT WEAN.
III
WHETHER THE MILITARY JUDGE ERRED IN ADMITTING THE OPINION TESTIMONY OF THE GOVERNMENT’S EXPERT WITNESSES ABSENT A SHOWING THAT THE UNDERLYING THERAPY TECHNIQUES THEY RELIED UPON IN REACHING THEIR CONCLUSIONS WERE RELIABLE.
IV
WHETHER THE ARMY COURT DECISION SHOULD BE REVERSED BECAUSE THAT COURT ABUSED ITS DISCRETION BY REJECTING THE FINDINGS OF THE EVIDEN-TIARY HEARING AND THE GOVERNMENT CONCESSION WITHOUT A LEGITIMATE RATIONALE,' AND BY FAILING TO ADDRESS THE ISSUES OF INEFFECTIVE ASSISTANCE OF COUNSEL IN ACCORDANCE WITH THIS COURT'S DECISION IN UNITED STATES V. WEAN, 37 MJ 286 (CMA 1993).
. See 41 MJ 213, 229 n. * (1994).
. The four questions posed by this Court are as follows:
(1) ''[W]hat were the reasons that defense counsel failed to object to admission of 'play therapy' evidence used by government witnesses in this case to conclude that a child had been sexually abused?’’
(2) "[Wjhat were the reasons that defense counsel failed to request funds for or the services of an expert witness in the preparation of appellant's defense?”
(3) "[Wjhat were the reasons for defense counsel’s statement during closing argument on sentencing ‘that an illness of the mind then compelled [appellant] to do these things?' "
(4) "[D]oes defense counsel's post-trial affidavit adequately address appellant's claim that defense counsel ignored his alibi defense and failed to call a potential key witness?”
United States v. Wean, 37 MJ at 288.
. At the time of appellant's trial, members were allowed to reconsider findings of guilty reached by them at any time before announcement of the sentence. RCM 924(a), Manual for Courts-Martial, United States (1994 ed.). Currently, however, only the military judge is allowed to reconsider findings before sentence is announced. Members may only reconsider findings up to the point when findings are announced. RCM 924(a), Manual, supra (1995 ed.); see Drafters’ Analysis of RCM 924, Manual, supra (1995 ed.) at A21-66.