This appeal revisits a question with which this court has struggled on numerous occasions: When can the State use evidence of other bad acts in a prosecution for sexual abuse. A jury convicted defendant John Mitchell of three counts of second-degree sexual abuse in violation of Iowa Code sections 709.1 and 709.3 (1997) after hearing, over his objection, the testimony of two other victims of his alleged abuse. The court of appeals affirmed, concluding the district court did not abuse its discretion in allowing this other bad-acts evidence. On further review, we now reverse and remand for a new trial.
I. Background Facts and Proceedings.
Mitchell was originally charged in a six-count trial information with three counts of second-degree sexual abuse and three counts of indecent contact with a child, in violation of Iowa Code section 709 .12. The sexual-abuse charges arose out of Mitchell’s relationship with twelve-year-old *297Amy,1 the daughter of Mitchell’s then-girlfriend. Mitchell lived with Amy and her .mother. Amy alleged that Mitchell first touched her breasts and vagina over her clothes. She claimed the abuse quickly escalated, with Mitchell touching her under her clothes and attempting to penetrate her with his penis.
Prior to trial, Mitchell moved to sever these three counts from the remaining counts of indecent contact with a child. The other three counts involved three different victims. Two of the victims were eight and thirteen-year-old sisters, Susanna and Karen, who were Amy’s friends. They both alleged that Mitchell touched their breasts and vaginas, both over and under their clothes, when they spent the night with Amy in the apartment shared with Mitchell.
The court ultimately severed the six counts into three groups. The first trial included the three charges of second-degree sexual abuse involving Amy, referred to as counts I-III. Trial two covered the two charges of indecent contact with Susanna and Karen, referred to as counts IV and V. Count VI, involving a third victim, was eventually dismissed.
The appeal before us concerns only Mitchell’s alleged abuse of Amy. Mitchell moved in limine to exclude any testimony from Susanna and Karen regarding the allegations forming the basis for counts IV and V. Mitchell contended that the girls’ testimony would only be relevant to show propensity, and in any event, its potential prejudice would substantially outweigh any probative value. The State countered that the testimony was relevant to bolster Amy’s credibility. Mitchell’s cross-examination of Amy accused her of fabricating the allegations to break up his relationship with her mother.
The district court allowed each girl’s testimony. It first determined that the evidence was relevant. It then concluded that the testimony’s probative value was not outweighed by possible unfair prejudice to the defendant, given the State’s need for the evidence, the fact that the acts in question were not more sensational than the acts Amy testified to, and the fact Mitchell challenged Amy’s credibility on cross-examination. Prior to the girls’ testimony, the court gave the following limiting instruction:
[Y]ou are cautioned and instructed that we are now going to hear some evidence that is offered for a very limited purpose. The purpose is proof of motive, opportunity, preparation, plan, or absence of mistake or accident by the Defendant. This evidence, and I will instruct you this way at the close of the case, is not being admitted as proof that the Defendant did or did not commit the acts Plaintiff alleges in this case and you must not convict based on this evidence. It is offered and only may be considered for the limited purpose of proving the truth or falsity of [Amy’s] testimony.
A similar limiting instruction was given with other instructions at the close of all the evidence.
A jury convicted Mitchell of all three counts of second-degree sexual abuse.2 We transferred his appeal to the court of appeals, which affirmed. Mitchell applied for further review in light of our recent opinion in State v. Castaneda, 621 N.W.2d *298435 (Iowa 2001), and we granted his application.
II. Issue on Appeal.
Mitchell’s appeal questions the reach of Iowa Rule of Evidence 404(b). This rule recognizes two fundamental principles: first, that evidence of other bad acts is relevant to whether a defendant committed the charged crime, and second, the reality that such evidence, while highly probative, is also highly prejudicial to a defendant. The rule attempts to balance these competing interests by presuming that “other crimes” evidence is inadmissible and placing the burden on the State to show the evidence is relevant for some purpose other than to show the defendant acted in a manner conforming to the other bad acts. See State v. Cott, 283 N.W.2d 324, 326 (Iowa 1979) (“The purpose of the rule is to exclude from the jury’s consideration evidence which has no relevancy except to show the defendant is a bad person and thus likely committed the crime in question.”).
Mitchell contends the State has not met its burden here. Although the State argued at trial, and urges on appeal, that Susanna’s and Karen’s testimony was relevant to corroborate Amy’s testimony and bolster her credibility, Mitchell asserts that their testimony is nothing more than propensity evidence. By its very nature, the tendered proof suggested that Susanna and Karen were testifying truthfully, that Mitchell had previously abused them, and that Mitchell acted in conformity with this behavior by abusing Amy. Thus, Mitchell argues, the jury was able to use testimony of other bad acts for the exact purpose for which it is not allowed under rule 404(b): to show Mitchell’s bad character in order to prove that he acted in conformity therewith.
III. Application of Iowa Rule of Evidence 404(b).
As noted above, Iowa Rule of Evidence 404(b) controls the admissibility of prior bad-acts evidence, stating:
Evidence of other crimes, wrongs, or acts is not admissible to prove the character of a person in order to show that he acted in conformity therewith. It may, however, be admissible for other purposes, such as proof of motive, opportunity, intent, preparation, plan, knowledge, identity, or absence of mistake or accident.
The alternative purposes listed in the rule are not exclusive. State v. Plaster, 424 N.W.2d 226, 228 (Iowa 1988). The important question is whether the disputed evidence is “relevant and material to some legitimate issue other than a general propensity to commit wrongful acts.” State v. Barrett, 401 N.W.2d 184, 187 (Iowa 1987); accord Castaneda, 621 N.W.2d at 440. If the challenged evidence is relevant to a legitimate issue in dispute, then it is prima facie admissible, regardless of any tendency to also establish a defendant’s bad character or propensity for committing bad acts. Plaster, 424 N.W.2d at 229.
Courts engage in a two-step process to determine whether such challenged evidence is admissible. First, the court must determine whether the evidence of other crimes or bad acts is relevant to a legitimate factual issue in dispute. State v. Bayles, 551 N.W.2d 600, 604 (Iowa 1996). If the evidence is deemed relevant to a legitimate factual issue in dispute, the court must then decide if its probative value is substantially outweighed by the danger of unfair prejudice to the defendant. Id.; Iowa R. Evid. 403. If the answer to the second question is yes, this overcomes the evidence’s prima facie ad*299missibility, and the evidence must be excluded. Bayles, 551 N.W.2d at 604.
We have stated that these decisions invoke the district court’s sound discretion but have recognized that that discretion is circumscribed as follows:
“It should be recognized, however, that this is not a discretion to depart from the principle that evidence of other crimes, having no substantial relevancy except to ground the inference that accused is a bad man and hence probably committed this crime, must be excluded. The leeway of discretion lies rather in the opposite direction, empowering the judge to exclude the other-crimes evidence, even when it has substantial independent relevancy, if in his judgment its probative value for this purpose is outweighed by the danger that it will stir such passion in the jury as to sweep them beyond a rational consideration of guilt or innocence of the crime on trial. Discretion implies not only leeway but responsibility. A decision clearly wrong on this question of balancing probative value against danger of prejudice will be corrected on appeal as an abuse of discretion.”
State v. Johnson, 224 N.W.2d 617, 621 (Iowa 1974) (quoting Edward W. Cleary, McCormick on Evidence § 190, at 453-54 (2d ed.1972)).
The State argues that the consequential fact in this case was Amy’s credibility. It distinguishes this issue of credibility from the general issue of credibility that attaches to every witness’s testimony by noting Mitchell’s implication, on cross-examination, that Amy fabricated her allegations to thwart Mitchell’s relationship with her mother. As in so many sexual abuse cases, Amy’s credibility was crucial to the State’s case because the evidence boiled down to “he-said” versus “she-said.”
As demonstrated here, the linchpin between the sisters’ testimony and Amy’s credibility is the premise that, if Mitchell abused Susanna and Karen, it was more likely that he abused Amy, and thus Amy was telling the truth. In other words, the only way the evidence could bolster Amy’s credibility was by proving that Mitchell was guilty of the crime. Consequently, the district court’s limiting instruction was self-nullifying.
Some of the same concerns involved in this case were present in Plaster, 424 N.W.2d at 228, in which we found prior bad-acts evidence to be relevant to the issue of consent in a sexual abuse case. That case, decided by a sharply divided court, is sufficiently distinguishable from the present case that it is not direct precedent standing in the way of our decision. The future applicability of Plaster in a factual situation similar to the one in that case now remains an open question.
The New Hampshire Supreme Court recently rejected an attempt to use other bad-acts testimony to bolster a victim’s credibility. See State v. Glodgett, 144 N.H. 687, 749 A.2d 283 (2000). We find the court’s reasoning persuasive and quote it here:
We reject the argument that the evidence of other sexual assaults was properly admitted to provide context to evaluate the complainant’s credibility in light of the defendant’s position that the assault never occurred. “Context, in this instance, is merely a synonym for propensity.” Absent a specific motive to lie that the evidence of uncharged sexual assaults, if believed, would rebut, the evidence is probative of the complainant’s credibility only if one concludes that the defendant committed the uncharged acts and infers that he committed the charged acts. “When, in this manner, an assumption based upon the defendant’s propensity toward certain action is the essential connection in the inferential chain supporting relevance, *300the evidence is inadmissible under Rule 404(b)
Id. at 289 (emphasis added) (quoting State v. Melcher, 140 N.H. 823, 678 A.2d 146, 150 (1996)). Other courts have reached similar conclusions. See People v. Sabin, 463 Mich. 43, 614 N.W.2d 888, 902 (2000) (rejecting use of prior-bad-acts testimony from another victim to bolster complainant’s credibility because not part of the “principal transaction” — the abuse of the victim of the charged acts); Evert v. Swick, 300 Mont. 427, 8 P.3d 773, 777 (2000) (testimony about prior bad acts from another victim not allowed under rule 404(b) when only reason offered is to bolster victim’s credibility and harm the defendant’s).
While we note that other courts have adopted a contrary view, e.g., Hardeman v. State, 247 Ga.App. 503, 544 S.E.2d 481, 485 (2001); Rigler v. State, 941 P.2d 734, 738 (Wyo.1997); Daniel v. State, 923 P.2d 728, 735 (Wyo.1996), we are convinced that doing so eviscerates the language and spirit of rule 404(b). A plea of not guilty in a criminal case places in issue the credibility of all the State’s witnesses. If the State is allowed to prevail on its theory that there is an independent relevancy to bad-acts evidence for credibility purposes, this doctrine could be invoked in nearly every criminal case.
Therefore, we hold the district court abused its discretion in admitting Susanna’s and Karen’s testimony concerning Mitchell’s prior bad acts because such testimony spoke to no legitimate fact besides Mitchell’s propensity to abuse young girls. Any other result would cause exclusion of prior-bad-acts evidence under 404(b) to become the exception rather than the rule.
IV. Conclusion and Disposition.
Because we conclude that the testimony of Susanna and Karen was not relevant to prove any legitimate fact other than Mitchell’s propensity to abuse young girls, we vacate the decision of the court of appeals, reverse the decision of the district court, and remand for a new trial. Our disposition of the case makes it unnecessary to address Mitchell’s ineffective-assistance-of-counsel claim.
DECISION OF COURT OF APPEALS VACATED; DISTRICT COURT JUDGMENT REVERSED AND REMANDED.
All justices concur except NEUMAN and TERNUS, JJ., who dissent. SNELL, S.J.,* participates in lieu of STREIT, J., who takes no part.. We have not used the victims’ true names in this opinion.
. In a second jury trial, Mitchell was also convicted of one count of indecent contact with a child, in violation of Iowa Code section 709.12 (1997), based on the charge involving Karen. This conviction has not been challenged, and thus forms no part of this appeal.