Appellee Dewayne Dee Dillard was convicted in the Sebastian County Circuit Court of two counts of first-degree sexual abuse and sentenced to eight years’ imprisonment. This court affirmed the convictions. See Dillard v. State, 333 Ark. 418, 971 S.W.2d 764 (1998). Dillard then filed a petition for postconviction relief, pursuant to Ark. R. Crim. P. 37, asserting that his trial counsel was ineffective for failing to interview and subpoena two witnesses who would have testified that one of the victims was untruthful. The circuit court granted the petition, and the State of Arkansas brings this appeal. Our jurisdiction is pursuant to Ark. Sup. Ct. R. 1-2(a) (7). We affirm.
Before we reach the merits of this appeal, we must first address Dillard’s argument that the State may not appeal from the grant of a Rule 37 petition. The State responds by contending that Rule 37 proceedings are civil in nature, and that this appeal is brought pursuant to Rule 2(a)(3) of the Arkansas Rules of Appellate Procedure—Civil (providing for appeal from an order granting a new trial). We agree with the State, as this court has repeatedly recognized that postconviction proceedings under Rule 37 are civil in nature. See, e.g., Cravey v. State, 306 Ark. 487, 815 S.W.2d 933 (1991); Mullins v. State, 303 Ark. 695, 799 S.W.2d 550 (1990) (per curiam); Brooks v. State, 303 Ark. 188, 792 S.W.2d 617 (1990) (per curiam). Indeed, this court has recently decided appeals by the State from grants of postconviction relief. See State v. Clemmons, 334 Ark. 440, 976 S.W.2d 923 (1998); State v. Her-red, 332 Ark. 241, 964 S.W.2d 391 (1998); State v. Slocum, 332 Ark. 207, 964 S.W.2d 388 (1998). Having thus determined that the appeal is properly before us, we turn to the merits of the case.
The record reveals that Dillard was convicted of two counts of first-degree sexual abuse involving two minor girls: S.S. and T.L. During the course of the trial, Dillard maintained that the girls were lying. The only witness called by the defense was Dillard himself. Dillard alleged in his petition that his trial counsel, John Settle, was ineffective for failing to interview and subpoena two potential defense witnesses: Samantha Bay, S.S.’s sister, and *574Jimmy Lambert, T.L.’s brother.1 At the hearing, Dillard testified that Bay would have testified that S.S. was not a truthful person, and that she did not believe S.S.’s accusations against Dillard. Dillard stated that Lambert was prepared to testify that S.S. had made up the allegations against Dillard.
Dillard maintained that he had informed Mr. Settle about these two witnesses. He stated that Mr. Settle assured him that he did not need any witnesses because he was going to rely primarily on the motion for severance of the two charges. He stated that Mr. Settle said that he was going to continue to rely on the severance motion up to the day of trial and beyond. He stated that Mr. Settle told him that there would be no problem in obtaining a severance through an appeal, and that he should not worry about it because he would be entitled to a new trial on appeal. The trial record reflects that the only witness called by the defense was Dillard himself.
Mr. Settle testified that he was aware that Bay might have held the opinion that S.S. had a problem telling the truth and had a reputation for dishonesty. He indicated that her testimony, as described in the Rule 37 petition, would have been helpful to the defense. He stated that he knew that Dillard’s trial was going to come down to a swearing match between Dillard and the two victims. Mr. Settle conceded, however, that he had never spoken with Bay about her potential testimony, and he maintained that he made the decision not to call her as a witness based upon his experience as a criminal defense attorney. He explained that he was aware that Bay hated Dillard, and that she had stated that she hoped he rotted in jail or was hanged by a rope. He stated that Bay was not someone he wanted to “let loose on the stand.” Mr. Settle agreed with Dillard’s counsel, however, that he could have avoided the unfavorable testimony by simply not asking Bay her opinion of Dillard.
Mr. Setde denied having any knowledge of Lambert’s testimony, which he stated would have been extremely helpful to the *575defense. He denied having ever been informed by Dillard of this witness’s existence. In contrast, Dillard’s father, Erbie Dillard, testified that he was present during a conference with Mr. Settle when Dillard tried to get him to call Lambert as a witness. Erbie Dillard stated that Mr. Settle indicated that he did not think they would need to call any witnesses.
The trial court granted Dillard’s petition on the ground that trial counsel was ineffective in regard to both witnesses’ testimony. The trial court found that counsel was ineffective for relying entirely on Samantha Bay’s statement to the authorities and failing to interview her and make an independent evaluation of her potential testimony. The trial court also found that counsel was ineffective for failing to interview and call Jimmy Lambert as a witness at Dillard’s trial. The trial court found further that Dillard’s trial had been prejudiced in these two instances. On appeal, the State raises three points for reversal. We find no merit and affirm the trial court’s order.
The criteria for assessing the effectiveness of counsel were enunciated by the Supreme Court in Strickland v. Washington, 466 U.S. 668 (1984). Strickland provides that when a convicted defendant complains of ineffective assistance of counsel, he must show (1) that counsel’s representation fell below an objective standard of reasonableness, and (2) that counsel’s deficient performance prejudiced his defense. See Clemmons, 334 Ark. 440, 976 S.W.2d 923; Thomas v. State, 330 Ark. 442, 954 S.W.2d 255 (1997). Unless the petitioner makes both showings, it cannot be said that the conviction resulted from a breakdown in the adversarial process that renders the result unreliable. Id. In other words, the petitioner must show that there is a reasonable probability that, but for counsel’s errors, the factfinder would have had a reasonable doubt respecting guilt in that the decision reached would have been different absent the errors. Id. A reasonable probability is a probability sufficient to undermine confidence in the outcome of the trial. Id. We will not reverse the trial court’s decision granting or denying postconviction relief unless it is clearly erroneous. Herred, 332 Ark. 241, 964 S.W.2d 391; Slocum, 332 Ark. 207, 964 S.W.2d 388. With this standard in mind, we discuss the points on appeal.
*576 I. Samantha Bay’s Testimony
The trial court found that Mr. Settle was ineffective for failing to interview Samantha Bay, S.S.’s sister, and make an independent evaluation of her potential testimony. The trial court found further that Bay’s testimony related to the credibility of one of the victims, S.S., to the extent that the victim was not a truthful person and could not be believed. The State argues that Mr. Settle’s decision not to call Bay as a witness was a matter of trial strategy, and that the decision not to interview her was reasonable because she was not going to be called as a defense witness. The State argues further that even if Mr. Settle’s decision not to interview Bay or call her as a witness was unreasonable, Dillard was not prejudiced because Bay would have been a mere character witness. We disagree.
The decision of whether or not to call a witness is generally a matter of trial strategy that is outside the purview of Rule 37. Catlett v. State, 331 Ark. 270, 962 S.W.2d 313 (1998); Helton v. State, 325 Ark. 140, 924 S.W.2d 239 (1996). Trial counsel must use his or her best judgment to determine which witnesses will be beneficia! to the client. Johnson v. State, 325 Ark. 44, 924 S.W.2d 233 (1996); Wicoff v. State, 321 Ark. 97, 900 S.W.2d 187 (1995). When assessing an attorney’s decision not to call a particular witness, it must be taken into account that the decision is largely a matter of professional judgment that experienced advocates could endlessly debate, and the fact that there was a witness or witnesses that could have offered testimony beneficial to the defense is not in itself proof of counsel’s ineffectiveness. Johnson, 325 Ark. 44, 924 S.W.2d 233. Nonetheless, such strategic decisions must still be supported by reasonable professional judgment pursuant to the standards set forth in Strickland, 466 U.S. 668. Wicoff, 321 Ark. 97, 900 S.W.2d 187.
In Wicoff, 321 Ark. 97, 900 S.W.2d 187, this court reversed the trial court’s denial of postconviction relief due to counsel’s errors in (1) failing to investigate the victims’ prior allegations of sexual abuse, and (2) fading to call the defendant’s grandmother, Ms. Dickerson, who would have testified that one of the young victims told her that she had fabricated the allegations against the *577defendant. Trial counsel explained that he had not called Ms. Dickerson because she was a relative of the defendant. This court rejected that explanation, holding:
[W]e cannot say that, under the facts presented, [trial counsel’s] decision was supported by reasonable professional judgment. At trial, the evidence against Wicoff was limited to the testimony of the tivo girls. The doctor who examined them testified that there was no physical evidence of sexual abuse, and that, in his opinion, the girls’ accounts of the incidents were reliable. As such, the credibility of the girls’ testimony was highly relevant, as their testimony was, in essence, the State’s only evidence against Wicoff. Ms. Dickerson’s testimony that the eleven-year-old had admitted to her that she had fabricated the incident would have served to impeach the minor’s credibility. Although Ms. Dickerson was available to testify to this conversation at trial, counsel for WicofF did not call her or any witnesses on his behalf. Counsel for Wicoff readily admitted, however, through his own testimony at the Rule 37 hearing, that Ms. Dickerson’s testimony “would have shed doubt on [Wicoffs] guilt [sic]. ”
Id. at 102, 900 S.W.2d at 189-90 (emphasis added). This court concluded that there was a reasonable probability that the outcome of the trial would have been different, but for counsel’s errors.
Here, Mr. Settle conceded that he was aware that Bay might have held that opinion that S.S. had a problem telling the truth and had a reputation for dishonesty. Yet he declined to interview her despite his knowledge that Dillard’s trial was going to come down to a swearing match between Dillard and the two victims. As such, his failure to interview the girl was unreasonable in light of the holding in Wicoff. Moreover, we concur with the trial court’s finding that Dillard’s trial was prejudiced by counsel’s error in this instance. It is undisputed that there was no physical evidence presented in support of the charges against him. Like the situation in Wicoff, the State’s case rested almost entirely on the testimony of the two young victims. Accordingly, the victims’ credibility, or lack thereof, was of crucial importance. Mr. Settles’s explanation that he did not call Bay because he was afraid that she would reveal her hatred for Dillard is unreasonable, as it clearly would have been to the defense’s advantage to have a wit*578ness who admittedly hated the defendant testify that the victim, the witness’s own sister, was not to be believed.
II. Jimmy Lambert’s Testimony
During the Rule 37 hearing, Lambert testified that a couple of weeks prior to the alleged incident, S.S. talked about being mad at Dillard and how she would call the cops and tell them that Dillard had raped her. Lambert stated that he told Dillard that he was willing to testify at the trial; however, Lambert stated that.he was never contacted by Mr. Settle regarding his testimony. Mr. Settle admitted that Lambert’s testimony would have been extremely helpful, but he maintained that he had not been aware of the testimony. The trial court was not persuaded by Mr. Settle’s claim, as it found that counsel was ineffective for failing to interview Lambert and for failing to call him as a defense witness.
The State makes no argument regarding trial counsel’s failure to interview Lambert and call him as a defense witness. Rather, the State asserts that Dillard was not prejudiced by trial counsel’s decision not to call Lambert because there was other evidence supporting his defense that S.S. had made up the allegations against him. This “other evidence” referred to by the State is the testimony of Detective Holohan, who testified that S.S. told him that she hated Dillard and that whenever she was mad at someone, she took it out on them. Holohan also testified that Dillard told him that S.S. was making up the allegations because she did not like him and was trying to undermine his relationship with her mother. The State contends that Lambert’s testimony would have been cumulative, and its omission was thus not prejudicial. We disagree that his testimony was merely cumulative.
As discussed in the previous point, the victims’ credibility, or lack thereof, was of crucial importance to the defense, as the State’s case against Dillard rested almost entirely on the testimony of the two girls. Jimmy Lambert is the brother of T.L., one of the victims in this case. Like Bay, Lambert’s relationship to one of the victims in this case lends credibility to his testimony, as it may have tended to undermine the entire case against Dillard. His testimony was therefore of chief importance to the defense, and *579would likely have carried more weight than the testimony offered by Dillard or Detective Holohan. Thus, for the reasons explained in Wicoff, 321 Ark. 97, 900 S.W.2d 187, we cannot say that the trial court erred in finding that Mr. Settle’s decision not to call Lambert as a defense witness was prejudicial to Dillard’s trial.
III. Prejudice to the Count Involving T.L.
For the final point for reversal, the State argues that, at a minimum, the trial court’s ruling must be reversed as to the charge involving the sexual abuse of T.L. The State submits that each of the errors alleged by Dillard pertained only to the credibility of S. S. and did not call into question T.L.’s credibility. Thus, the State asserts, Dillard should only receive a new trial on the count involving S.S. Although we agree with the State that the proffered testimony from both witnesses involved the lack of veracity of only one of the victims, we cannot say that the proffered testimony would not have otherwise impacted the charges involving T. L.
Dillard was tried for one count of rape against S.S. and one count of first-degree sexual abuse against T.L. The jury found Dillard guilty of engaging in first-degree sexual abuse with both victims. The only issue pursued on appeal was whether the trial court erred in refusing to grant Dillard’s motion to sever the charges. This court concluded that the allegations made by the girls were so similar that each victim’s testimony would have been relevant and admissible in separate trials:
The thirty-five-year-old Appellant obtained access to both young female victims through his relationships with their respective parents; he stayed with S.S.’s mother off and on, and visited T.L.’s home frequently during the time frame at issue. Additionally, T.L. had lived with S.S.’s mother and Appellant for a brief period after T.L.’s house had burned. As such, we conclude that Appellant formed intimate relationships with both victims. The similarity of the manner in which Appellant committed the acts is remarkable. Both victims testified that Appellant made sexual advances toward them and grabbed their breasts on numerous occasions. S.S. testified that Appellant raped her after she had refused his ongoing demands for sexual intercourse. T.L. testi*580fled that Appellant threatened to rape her if she would not consent to sex. Both victims testified that he told them that sex would not hurt and would only take about three minutes. Moreover, the overlapping time frame of June 19, 1996, through August 2, 1996, also weighs in favor of proving Appellant’s intent, scheme, or plan.
Dillard, 333 Ark. 418, 426, 971 S.W.2d 764, 768.
To prevail on a claim of ineffective assistance of counsel, the petitioner must only show that there is a reasonable probability that, but for counsel’s errors, the factfinder would have had a reasonable doubt respecting guilt, i.e., the decision reached would have been different absent counsel’s errors. Thomas, 330 Ark. 442, 954 S.W.2d 255. Because the testimony of the girls was so similar, we cannot say that the trial court clearly erred in concluding that Dillard’s trial was prejudiced by counsel’s errors. Accordingly, the judgment of the trial court is affirmed.