This case involves the issue of whether admitting evidence of prior misconduct as a common scheme or plan under Evidence Rule (ER) 404(b) requires that the evidence show a unique or atypical method of *13committing the crime, or merely show sufficient similarities to suggest a common scheme or plan. In a prior case, Division Two of the Court of Appeals held that the common scheme or plan exception requires the similarities to be atypical or unique to the way the crime is usually committed. In the present case, Division One disagreed and held that the exception requires only substantial similarities between the acts, which are naturally explained by a common scheme or plan. We agree with Division One and hold that to admit evidence of prior bad acts as evidence of a common scheme or plan under ER 404(b), the trial court need only find that the prior bad acts show a pattern or plan with marked similarities to the facts in the case before it. We affirm Louis A. DeVincentis’ conviction and the Court of Appeals opinion in this case.
FACTUAL AND PROCEDURAL HISTORY
In the summer of 1998, K.S. was 12 years old. Her friend, C.K., lived next door to DeVincentis. That summer, De-Vincentis offered the girls money to mow his lawn. After checking with their mothers, the girls accepted and began to mow his lawn together about once a week.
Sometime during September of 1998, DeVincentis asked K.S. and C.K. to clean his house. After the girls cleaned together once, DeVincentis asked only KS. to come back. While K.S. cleaned, DeVincentis would be present in the house, wearing only a g-string or bikini underwear. DeVincentis would acknowledge that he was in a state of undress by saying something to the effect of “I hope you don’t mind.” Report of Proceedings (RP) at 908. No one else would be present during these times.
One day at the end of October, K.S. arrived at De-Vincentis’ house to clean. DeVincentis told her that he was sore from exercising and asked her to give him a massage. Wearing only bikini underwear, DeVincentis lay on the couch on his stomach and K.S. massaged his back. After the massage, K.S. felt uncomfortable and left. DeVincentis told her not to tell anyone or she would be in trouble.
*14A few weeks later, K.S. returned to clean DeVincentis’ house at his request. When she reached the bedroom, De-Vincentis was there, again wearing only a g-string. He asked K.S. to massage him. He lay on the floor and told K.S. to take off her clothes, which she did. As K.S. massaged DeVincentis’ back, he took off his g-string and asked her to massage his buttocks. DeVincentis then directed K.S. to massage his legs. Then, DeVincentis told K.S. to lie down, and he massaged her back, buttocks, and legs. DeVincentis then lay on his back and asked K.S. to massage his stomach and eventually his penis until he ejaculated. K.S. testified that DeVincentis then massaged her chest. He then touched inside her vagina, hurting her. K.S. testified that she was scared during this and wanted to go home. K.S. told him she had to go home, and as she left, he again told her not to tell anyone.
K.S. returned to DeVincentis’ home again in November. DeVincentis wore a g-string. After she finished cleaning, K.S. found DeVincentis in his bedroom. DeVincentis asked K.S. if she wanted to give him another massage. She testified that she was scared but said yes. K.S. testified that she was afraid that DeVincentis would not allow her to refuse. He told K.S. to remove her clothes, and the massaging and sexual contact proceeded as it had the previous time. Afterward, DeVincentis again warned K.S. not to tell anyone or she would be in trouble.
K.S. eventually told her mother about these events. Her mother reported the events to the police in January 1999. DeVincentis was charged with one count of second degree rape of a child and one count of second degree child molestation. Clerk’s Papers (CP) at 1.
Evidence of Prior Bad Acts
Prior to trial, the State sought admission of evidence that DeVincentis had been convicted of crimes involving sexual *15misconduct with young adolescent girls in New York several years before, including those with Y.C., which led to a 1983 conviction for sexual abuse.
At a hearing, V.C. testified that in 1983 she was 10 years old and the best friend of DeVincentis’ daughter. V.C. spent three or four evenings a week at DeVincentis’ home and he was usually present, wearing nothing but bikini or g-string underwear. V.C. testified that the sight of DeVincentis in small underwear became normal to her. After a birthday party for DeVincentis’ daughter, DeVincentis took V.C. to his bedroom and showed her pictures of naked people. He asked her if she had seen a penis before. DeVincentis was wearing bikini briefs, and he asked V.C. if it bothered her that he was dressed like that.
On another occasion, DeVincentis had V.C. sit on his home rowing machine. DeVincentis sat behind her and she felt what she now understands to be his erection pressing against her back. On this occasion he also “put[ ] his hand in [her] private areas and press [ed] down on them hard.” RP at 509. DeVincentis then showed her another exercise machine and touched her chest and private area and pressed his erection against her back. On another occasion, DeVincentis demanded that V.C. try on transparent, mesh-like clothing. DeVincentis stored books and magazines with pictures of nude people throughout his house where his daughter and V.C. could find them. Once he offered V.C. $10 to pose for a nude picture like one in a magazine.
V.C. testified that she had memory flashes of being naked with DeVincentis in his bedroom and him asking for back massages. V.C. also testified that she had memory flashes of DeVincentis putting his erection on her and in her mouth and that he ejaculated on her.
The State argued that these prior bad acts were sufficiently similar to the charged crimes to show a common scheme or plan to molest young girls and were therefore admissible under State v. Lough, 125 Wn.2d 847, 889 P.2d 487 (1995). Over DeVincentis’ objection, the trial court agreed and admitted the testimony of V.C. The court *16excluded other evidence of DeVincentis’ sexual misconduct with other prior victims, but ruled that the testimony of V.C. “established] sufficient similarity in the modus operandi and the location to support a common scheme or plan.” CP at 10; RP at 226-27.
In its written findings, the trial court noted that in DeVincentis’ conduct with both V.C. and K.S., he wore only bikini or g-string underwear, giving the impression that it was normal, and asked both girls if they minded his lack of clothes. He asked both girls to remove their clothes. He asked both for massages and to masturbate him until he ejaculated. He asked both girls not to tell. The trial court ruled that DeVincentis’ lack of clothing showed a design or plan to add a sense of normalcy to his behavior and to gain the trust of the girls by desensitizing them to his nudity, thereby making “it easier to move from nudity to physical skin-to-skin touch and sexual behavior.” CP at 10. The trial court found the evidence of this plan relevant to whether DeVincentis had sexual contact with K.S. In a bench trial, DeVincentis was convicted as charged.
Court of Appeals Opinion
Division One of the Court of Appeals agreed with the trial court that V.C.’s testimony was admissible under Lough. Rejecting an interpretation of Lough by Division Two requiring the similarities to be unique or atypical,1 Division One held that there were sufficient similarities between V.C.’s testimony and that of K.S. to support finding a common scheme or plan. State v. DeVincentis, 112 Wn. App. 152, 162, 47 P.3d 606 (2002). “A rational fact finder could well conclude that DeVincentis used a particular strategy to win each girl’s cooperation, over time, in her own victimization.” DeVincentis, 112 Wn. App. at 162-63.
*17ANALYSIS
Divisional Split on Requirements for a Common Scheme or Plan
At issue here is the proper factual foundation for the admission of evidence of prior bad acts as proof of a common scheme or plan under ER 404(b). This court reviews the correct interpretation of an evidentiary rule de novo as a question of law. See State v. Walker, 136 Wn.2d 767, 771-72, 966 P.2d 883 (1998). Once the rule is correctly interpreted, the trial court’s decision to admit or exclude evidence is reviewed for an abuse of discretion. See Lough, 125 Wn.2d at 856.
A trial court must always begin with the presumption that evidence of prior bad acts is inadmissible. ER 404(b) prohibits admission of evidence to prove a defendant has a criminal propensity. The rule states that
[e]vidence of other crimes, wrongs, or acts is not admissible to prove the character of a person in order to show action 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.
ER 404(b).
The State must meet a substantial burden when attempting to bring in evidence of prior bad acts under one of the exceptions to this general prohibition. We have established the analysis for admission of evidence of prior bad acts to prove a common scheme or plan in Lough, 125 Wn.2d 847. The prior acts must be “(1) proved by a preponderance of the evidence, (2) admitted for the purpose of proving a common plan or scheme, (3) relevant to prove an element of the crime charged or to rebut a defense, and (4) more probative than prejudicial.” Lough, 125 Wn.2d at 852.
The third step is at issue in this case. Because the issue is whether the crime occurred, the existence of a design to fulfill sexual compulsions evidenced by a pattern *18of past behavior is probative. Therefore, prior bad acts may be admitted to show a plan or design if they satisfy the substantial threshold articulated in Lough. We are asked to determine whether that threshold from Lough allows prior acts to be admitted which are similar to the charged crime but are not unique or uncommon.
As indicated, the divisions of our Court of Appeals have differed on this issue. Division Two has interpreted Lough to require that the similarities between the prior bad acts and the facts of the charged crime must be unique or uncommon to the way the crime is typically committed. State v. Dewey, 93 Wn. App. 50, 58, 966 P.2d 414 (1998). Dewey examined the facts in Lough and concluded that “[t]he common features pointed to by the court as evidence of the plan were features not common to most rapes .. ..” Dewey, 93 Wn. App. at 56. Dewey further reasoned that if the common features were not “unique or uncommon to the way in which the crimes are typically committed [,] ... it is difficult to imagine a case in which evidence of prior misconduct would be excluded.” Dewey, 93 Wn. App. at 57.2
In the present case, Division One disagreed and held that uniqueness was not a requirement for admitting prior bad acts as evidence of a common scheme or plan. Division One interpreted Lough to require that the prior bad acts be similar enough to be naturally explained as individual manifestations of an identifiable plan. DeVincentis, 112 Wn. App. at 159-61. Division One agreed, as do we, that “caution is called for in application of the common scheme or plan exception as defined in Lough.” DeVincentis, 112 Wn. App. at 159. Random similarities are not enough. Division One reasoned, however, that by requiring that the similarities be unique or uncommon, Division Two “confuses the common scheme or plan exception to ER 404(b) with the modus operandi exception,” which is used to prove the identity of the perpetrator, not whether the charged crime occurred. *19DeVincentis, 112 Wn. App. at 159. We agree with Division One’s interpretation.
In Lough we recognized two types of evidence of a common scheme or plan admissible under ER 404(b). The first type involves multiple crimes that constitute parts of a larger, overarching criminal plan in which the prior acts are causally related to the crime charged. An example of this type is a prior theft of a tool or weapon, which is used to perpetrate the subsequent charged crime, such as a burglary. The issue in Lough was the admissibility of evidence of a second type of common scheme or plan, which involves prior acts as evidence of a single plan used repeatedly to commit separate, but very similar, crimes. We held that evidence of this second type of plan may be admissible if the State establishes a sufficiently high level of similarity:
To establish common design or plan, for the purposes of ER 404(b), the evidence of prior conduct must demonstrate not merely similarity in results, but such occurrence of common features that the various acts are naturally to be explained as caused by a general plan of which the charged crime and the prior misconduct are the individual manifestations.
Lough, 125 Wn.2d at 860.
Although this burden is substantial, our holding focused on the similarity between the prior acts and the charged crime rather than the uniqueness of the individual acts. Lough looked to other jurisdictions for guidance in reaching this conclusion. We first looked to a California case, People v. Ewoldt, 7 Cal. 4th 380, 867 P.2d 757, 27 Cal. Rptr. 2d 646 (1994). Lough, 125 Wn.2d at 855-56. We quoted language from Ewoldt in formulating our analysis. This language includes the requirement that “the defendant committed ‘markedly similar acts of misconduct against similar victims under similar circumstances.’ ” Lough, 125 Wn.2d at 856 (quoting Ewoldt, 7 Cal. 4th at 399). Also, the acts must have “ ‘such a concurrence of common features that the various acts are naturally to be explained as caused by a general plan of which they are the individual manifesta*20tions.’ ” Lough, 125 Wn.2d at 856 (quoting Ewoldt, 7 Cal. 4th at 402).
We also looked to a case from the Minnesota Supreme Court, State v. Wermerskirchen, 497 N.W.2d 235 (Minn.1993). Wermerskirchen involved whether to admit prior acts of sexual misconduct under Minnesota’s own ER 404(b), which is substantially similar to Washington’s. Wermerskirchen, 497 N.W.2d at 236, 239. “The Minnesota high court explained that. . . the preferred approach is for the trial court to focus on the closeness of the relationship between the other misconduct and the charged crimes in terms of time, place and modus operandi.” Lough, 125 Wn.2d at 858. Although a unique modus operandi is one factor to consider, the crux of the inquiries in Ewoldt and Wermerskirchen is similarity, not uniqueness.
We emphasize that the degree of similarity for the admission of evidence of a common scheme or plan must be substantial. Wigmore’s treatise on evidence supports this conclusion. When the existence of the criminal act is at issue, evidence of substantially similar features between a prior act and the disputed act is relevant. Wigmore explains, however, that admission of this kind of evidence requires more than merely similar results. Wigmore gives the following example to illustrate when prior acts are admissible as common scheme or plan evidence:
So, on a charge of assault with intent to rape, where the intent alone is disputed, a prior assault on the previous day upon the same woman, or even upon another member of her family, might have probative value; but if the assault itself is disputed, and the defendant attempts, for example, to show an alibi, the same facts might be of little or no value, and it might be necessary to go further and to show (for example) that the defendant on the same day, with a confederate guarding the house, assaulted other women in the same family who escaped, leaving the complainant as the only woman accessible to him for his purpose.
2 John Henry Wigmore, Evidence § 304, at 249 (James H. Chadbourn rev. ed. 1979) (emphasis added). This height*21ened level of similarity does not require that the evidence of common features show a unique method of committing the crime.
Division Two acknowledged that Lough did not expressly require uniqueness but conducted its own analysis of the facts of Lough and determined that the features similar to the prior acts and the charged crime were uncommon to most rapes. Dewey, 93 Wn. App. at 56. Dewey reflects a misreading of Lough because our analysis in Lough requires similarity of the acts, not uniqueness.
The Dewey court also confused the common scheme or plan exception under ER 404(b) with the unique modus operandi exception, as Division One recognized. Evidence of unique modus operandi is relevant when the focus of the inquiry is the identity of the perpetrator, not whether the charged crime occurred. As we have recently established, when identity is at issue, the degree of similarity must be at the highest level and the commonalities must be unique because the crimes must have been committed in a manner to serve as an identifiable signature. State v. Vy Thang, 145 Wn.2d 630, 643, 41 P.3d 1159 (2002). In contrast, the issue in the present case was not the identity of the perpetrator, but whether the crime occurred. Although a unique method of committing the bad acts is a potential factor in determining similarity, uniqueness is not required.
In sum, admission of evidence of a common scheme or plan requires substantial similarity between the prior bad acts and the charged crime. Such evidence is relevant when the existence of the crime is at issue. Sufficient similarity is reached only when the trial court determines that the “various acts are naturally to be explained as caused by a general plan . . . .” Lough, 125 Wn.2d at 860. Applying this analysis to this case, in order to admit V.C.’s testimony, the trial court had to find a level of similarity between DeVincentis’ acts against V.C. and those against K.S., such that the two incidents were naturally explained as individual manifestations of a general plan.
*22In a lengthy oral ruling supplementing the written findings and conclusions, the trial court applied the four-part analysis required by Lough. The trial court evaluated whether V.C.’s testimony indicated that the prior acts were (1) proved by a preponderance of the evidence, (2) offered for the purpose of proving a common plan or scheme, (3) relevant to prove an element of the crime charged or to rebut a defense, and (4) more probative than prejudicial.
The trial court noted that the first step was not in dispute because the conduct with V.C. was the basis of a prior conviction. The trial court also noted that the conduct with V.C. was offered to prove a common scheme or plan. The court ruled that V.C.’s testimony was “relevant to show that the act was committed, a corpus delicti purpose . . . where the defendant denied the event.” RP at 220-21. The trial court explained that “the evidence involving [V.C.] is relevant to show that the defendant had devised a scheme to get to know young people through a safe channel, such as a friend of his daughter, or ... as a friend of the next-door neighbor girl. . . .” RP at 221. This led to “greater familiarity occurring in his own home. .. .” This plan allowed DeVincentis to bring the children into “an apparently safe but actually unsafe and isolated environment so that he could pursue his compulsion to have sexual contact with these . . . prepubescent or pubescent girls.” RP at 221. The girls were both between 10 and 13 years old.
Other similarities that the trial court noted included walking around his house in an unusual piece of clothing— bikini or g-string underwear. On both occasions, the trial court found that DeVincentis “indicated he . . . intended by the casual wearing of almost no clothes to reduce the children’s natural discomfort or negative reaction to such behavior. . . .” RP at 222. With both girls, DeVincentis “asked for a massage or gave [a] massage, asked or directed the child to a secluded spot such as a bedroom, directed or asked that clothes be taken off. . . .” RP at 223. Finally, in both instances, he had the girls masturbate him until climax. The trial court concluded that this was relevant to support K.S.’s testimony.
*23The trial court then balanced the probative value against the prejudicial effect. The trial court recognized that substantial probative value is needed to outweigh the prejudicial effect of ER 404(b) evidence. Oral argument on this issue before the trial court reflected that court’s understanding of relevant factors used in balancing, such as the age of the victim, the need for the evidence, the secrecy surrounding sex abuse offenses, “[t]he vulnerability of the victims, the absence of physical proof of the crime, degree of public opprobrium associated with the accusation,... [and the] general lack of confidence in the ability of a jury to assess the credibility of child witnesses.” RP at 130. After balancing these factors, the trial court found that the testimony of V.C. was the main evidence corroborating the testimony of K.S. No less inflammatory documentation or corroboration that the crime occurred was available. Although K.S. was old enough to clearly testify, on balance the probative value of V.C.’s testimony outweighed the prejudicial effect.
The trial court discussed a limiting instruction, despite the fact that it was a bench trial, reflecting the court’s understanding of the limited purpose for which it would use V.C.’s testimony.3 In closing argument, DeVincentis’ counsel further discussed the limited use of V.C.’s testimony, while acknowledging that judges are more able to limit their use of ER 404(b) evidence than are juries.
When this analysis is scrupulously applied by the trial court, it effectively prohibits mere propensity evidence. The present case exemplifies this. Although the State moved to admit several other prior occasions of sexual abuse of girls by DeVincentis, the trial court rejected them for lack of similarity with the charged crimes against K.S.
In sum, the trial court did not abuse its discretion when it admitted V.C.’s testimony under ER 404(b) because it *24meticulously applied the law and had tenable grounds and reasons for its decision.
Washington Association of Criminal Defense Lawyers’ Arguments
Amicus curiae, Washington Association of Criminal Defense Lawyers (WACDL), argues that Lough should be reversed. WACDL argues that Lough fails to give due consideration to the prejudice inherent in prior sex crimes, which leads to their unavoidable use as propensity evidence. WACDL argues that the trial court should focus on the defendant’s state of mind instead of focusing on the similarity of the prior acts when determining whether to admit prior acts as evidence of a common scheme or plan. Quoting from our opinion in State v. Saltarelli, WACDL maintains that:
“ ‘[w]hen deciding the issue of guilt or innocence in sex cases, where prejudice has reached its loftiest peak,. . . courts have been most liberal in announcing and fostering a nebulous exception, offering scant attention to inherent possibilities of prejudice. Just when protection is most needed, the rules collapse.’ ”
Br. of WACDL at 9 (quoting State v. Saltarelli, 98 Wn.2d 358, 364, 655 P.2d 697 (1982) (quoting M.C. Slough & J. William Knightly, Other Vices, Other Crimes, 41 Iowa L. Rev. 325, 334 (1956))). WACDL asserts that focusing on mental state, rather than similarity, will somehow protect against this danger.
Saltarelli emphasized the importance of a trial court determining relevance and balancing the prejudicial value against the probative effect of any prior bad act evidence brought in under ER 404(b). Saltarelli, 98 Wn.2d at 362. Weighing the prejudicial effect is especially important for cases involving sexual crimes, as the above quotation explains. Saltarelli, 98 Wn.2d at 363. In Lough, however, we emphasized a “common sense” approach for the use of prior bad acts to show a plan, especially in cases of sexual crimes, *25because the doing of the act is often difficult to prove. We quoted with approval another section of Wigmore’s treatise on evidence:
“The committing of a single previous rape, or rape attempt, upon another woman may not in itself indicate such a design .... Nevertheless, a single previous act, even upon another woman, may, with other circumstances, give strong indication of a design (not a disposition) to rape ....
“Courts have shown altogether too much hesitation in receiving such evidence. Even when rigorously excluded from any bearing it may have upon character ... , it may carry with it great significance as to a specific design or plan of rape. . . . There is room for much more common sense than appears in the majority of the rulings.”
Lough, 125 Wn.2d at 858-59 (quoting Wigmore, supra, § 357, at 335-42).
Our reliance on this passage reflects our concern that trial courts give special consideration to the probative value of such evidence when balancing against the prejudicial effect of such evidence, especially when corroborating evidence is not available. Saltarelli is consistent with both Lough and the Court of Appeals in the present case. In Saltarelli, we concluded that the evidence was improperly admitted because the trial court did not adequately balance probative value against prejudicial effect. Saltarelli, 98 Wn.2d at 363. In contrast, the trial court in the present case both articulated the correct standard for admitting common scheme or plan evidence and conducted a meticulous balancing of probative value and prejudicial effect, mindful of our reasoning in Lough. We decline to overrule Lough.
CONCLUSION
We affirm DeVincentis’ conviction and the Court of Appeals opinion in this case.
Alexander, C.J., and Madsen, Ireland, Bridge, Owens, and Fairhurst, JJ., concur.State v. Dewey, 93 Wn. App. 50, 966 P.2d 414 (1998).
Division Three agrees with Dewey. State v. Griswold, 98 Wn. App. 817, 825, 991 P.2d 657 (2000).
When ER 404(b) evidence is admitted, the trial court should instruct the jury of the limited purpose of such evidence. State v. Saltarelli, 98 Wn.2d 358, 362, 655 P.2d 697 (1982). The request for a limiting instruction must be made by the complaining party. See State v. Hess, 86 Wn.2d 51, 52, 541 P.2d 1222 (1975).