Skip to content
Search resources Open menu

Forensic Topics

Recidivism and Risk Assessment for Women Who Sexually Offend

Published

Snap Shot

  • The question: How often do women convicted of sexual offenses commit new sexual offenses, and how can an evaluator assess an individual woman’s risk when the standard tools were built for men?
  • What the research shows: Across 14 studies of 4,208 women, 3.1% (131 women) were known to have committed a new sexual offense over follow-ups that varied (2.4% within five years; the prediction interval across studies ran from near zero to about 11%), while about 30% had some new criminal charge or conviction; these are detected offenses, which undercount offending, and they describe a group rather than any one woman. The Static-99R did not predict sexual recidivism in a sample of 739 Texas women, and its coding rules and the leading adult guideline exclude women.
  • What it does not show: The figures count detected offenses only, and offenses by women appear to be reported to police less often than offenses by men. No tool has been adequately validated to estimate sexual recidivism risk for women (two newer instruments have shown preliminary results with very few female recidivists), the risk factors that have been identified rest on very few recidivists, and group base rates cannot show that a particular woman will or will not reoffend.
  • The law: Risk testimony is subject to Fed. R. Evid. 702, Mil. R. Evid. 702 and Ariz. R. Evid. 702. Civil commitment statutes, including A.R.S. § 36-3701 and 18 U.S.C. § 4248, apply to women as well as men and require a finding that the person is likely to engage in sexual violence or would have serious difficulty refraining from it.
  • Issues for each side: Prosecutors may point to research showing that some women do reoffend, that rates vary widely and that a few markers point to higher risk, but should not offer male instruments or norms; defense counsel may point to the low base rates and the absence of validated tools, but those cannot show that a particular woman poses no risk.

Courts, probation departments, parole boards and correctional systems are routinely asked to estimate the risk that a person convicted of a sexual offense will offend again. For men, those estimates rest on large recidivism studies and on validated instruments such as the Static-99R and STABLE-2007. For women who sexually offend, the research base is far smaller, the observed rates of new sexual offenses are far lower, and no instrument has been adequately validated to predict sexual recidivism. Evaluators and courts nonetheless face the same questions about women as about men: sentencing, supervision, registration level, treatment and, in some jurisdictions, civil commitment.

Two premises frame this paper. First, sexual abuse committed by women causes real harm. Victims of women describe depression, substance abuse, self-injury, relationship difficulties and problems with sexuality similar to those described by victims of men (Cortoni et al., 2017; Saradjian, 2010), and the low recidivism figures summarized below say nothing about the seriousness of an index offense or its effect on a victim. Second, a low group rate is not a zero rate. Some women do reoffend sexually, and the research gives limited guidance on how to identify them.

This paper summarizes the recidivism research, why tools developed for men are not used with women, the approaches recommended in their place, and the governing law. The prevalence of offending by women and the reasons it is underreported are addressed in “Women Who Sexually Offend: Prevalence and Underreporting,” and the use of risk instruments with other groups for whom they were not designed is addressed in “Risk Assessment Tools Outside Their Normative Samples.”

How Often Women Reoffend

The first meta-analysis of sexual recidivism among women combined 10 studies of 2,490 women from Canada, the United States, the United Kingdom, Australia and the Netherlands, with an average follow-up of 6.5 years (Cortoni et al., 2010). Recidivism was defined as a new charge, conviction or incarceration. Depending on the statistical model, the estimated rate of sexual recidivism ranged from about 1% to 3%; the rate of violent recidivism (including sexual offenses) ranged from about 4% to 8%; and the rate of any new offense ranged from about 19% to 24%. One Texas study, which counted any offense requiring registration, including prostitution-related and kidnapping offenses, as sexual recidivism, was a statistical outlier; with it removed, the sexual recidivism estimates were about 1%. By comparison, the authors cited five-year rates for men of about 14% for sexual recidivism, 25% for violent recidivism and 37% for any recidivism, drawn from a different, much larger set of studies.

The largest single study in that review followed 1,466 women convicted of registerable sexual offenses in New York State between 1986 and 2006 (Sandler & Freeman, 2009). Recidivism was measured by rearrest, and follow-up began at conviction rather than release, because most of the women received probation or short jail terms. Among women with at least five years of follow-up, 1.8% were rearrested for a sexual offense, 5.2% for a violent felony and 26.6% for any offense; over the whole study, 2.2% were rearrested for a sexual offense and 28 of those 32 women were reconvicted of one. In her chapter on assessment, Cortoni (2010) reanalyzed this sample and noted that 79 of the women had been convicted of offenses involving the prostitution of minors rather than contact sexual offenses. Ten of the 32 sexual recidivists came from that small group, whose rearrest rate for a new sexual offense was about 13%, compared with about 1.2% among the remaining women; Cortoni reported that the new offenses in that group were themselves prostitution-related. Offenses involving the prostitution of minors harm children, but in New York, as in many U.S. jurisdictions, the registered “sexual offender” category therefore covered two groups with very different histories and rates.

The most recent meta-analysis, published by the same research group in 2026, updated the evidence with 14 studies of 4,208 women, 11 of them from the United States, with publication dates from 1998 to 2023 (Hanson et al., 2026). It excluded adult sex-trade offenses but retained offenses involving the prostitution of children. Overall, 3.1% of the women were known to have sexually reoffended. The rate within the first five years was 2.4% (64 of 2,642 women), and between years five and ten it was 1.1% among women who had remained offense-free for five years, although only two studies reported that later period. The meta-analytic average was 2.8%, but the variation across studies was large, with a prediction interval from near zero to about 11%. Violent recidivism was 7.8% and any recidivism 30.1%. The authors noted that where the same study reported rates for both sexes, men’s sexual recidivism was higher in every sample but one. They also noted that the women’s rates were comparable to the rate of sexual offending among people with nonsexual convictions and men in the general population.

The highest rate cuts the other way. In a Nebraska sample of 57 women, 17.5% were charged with a new sexual offense, and when child-welfare reports were added to criminal charges, 16 of the 57 women (28%) had some later report of sexually inappropriate behavior (Bader et al., 2010). The sample was small, but the study shows that official charges can miss conduct that child-protection agencies record.

Online offending has received less study. In an Australian sample of 254 females processed by police for child sexual offending, the 116 known for online offenses were more often under 18 and more often had a co-offender than those known for offline offenses, but after control for other risk factors the groups did not differ in later police contact for any, violent or nonviolent offending (Fortunato et al., 2026). Sexual reoffending was not reported separately.

Why the Male Tools Do Not Transfer

The Static-99R is an actuarial instrument developed and normed on adult men. Its coding rules state that it is not recommended for females (Phenix et al., 2017), and the current guideline of the Association for the Treatment and Prevention of Sexual Abuse (ATSA) is limited by its own terms to men aged 18 and older. ATSA describes women who have committed sexually abusive behavior as a distinct population whose risk and protective factors need not parallel those of men (Association for the Treatment and Prevention of Sexual Abuse [ATSA], 2025). The STABLE-2007 was developed in a prospective study of men on community supervision; of the 997 people in that study, six were women (Cortoni et al., 2010).

Cortoni (2010) gave two reasons why tools developed for men should not be used with women. First, the probability estimates attached to male scores are calibrated to men’s base rates, so applying them to women, whose detected sexual recidivism is several times lower, would be expected to overestimate risk. That is an inference from base rates, not a tested result; the one direct test, in 739 women, found that Static-99R scores did not predict sexual recidivism (Marshall et al., 2021). Second, the items were selected because they predict reoffending among men, and there was no evidence that the same items, such as victim type, predict sexual recidivism among women. Cortoni et al. (2010) reached the same conclusion from the base rates alone.

That prediction has since been tested directly. In a sample of 739 women convicted of sexual offenses in Texas, the Static-99R total score was not significantly related to sexual recidivism; of the individual items, only prior sexual offenses and noncontact sexual offenses were related to it (Marshall et al., 2021). The authors concluded that the instrument is not suitable for women. It is the strongest evidence on the question but involved one state, official records and few sexual recidivists.

The same caution applies to sexual interest. Deviant sexual arousal and fantasy are part of the offending of some women, but Cortoni (2010) noted that little is known about their role and cautioned against applying knowledge about men to women. A review of methods for assessing paraphilic interests in women who sexually offend found little evidence that women’s genital responses indicate their sexual interests and no validated measure for this population (Bouchard et al., 2019).

What Evaluators Use Instead

General recidivism tools

Because women who have sexually offended are far more likely to reoffend nonsexually than sexually, Cortoni (2010) and Cortoni et al. (2010) recommended that evaluators use instruments validated for general and violent recidivism among women, such as the Level of Service Inventory–Revised (LSI-R). A meta-analysis of 27 effect sizes from samples totaling 14,737 female offenders found that LSI-R scores were related to recidivism among women about as strongly as among men (Smith et al., 2009). Those samples were general offenders, not women convicted of sexual offenses. Gender-responsive research has added factors, such as parental stress, relationship dysfunction, victimization, mental health and family support, that improved prediction beyond gender-neutral instruments in several samples of women offenders (Van Voorhis et al., 2010). In a sample of 225 women released after sexual offenses, followed for about five years, gender-specific factors such as mental illness symptoms and victimization history were related to reoffending (Marshall & Miller, 2019). A general tool answers a different question from the one a court may be asking: it estimates the likelihood of any new offense, not a new sexual offense.

Base rates and structured clinical judgment

For the specific question of sexual recidivism, the published recommendations are to begin with the base rate and adjust it in light of risk and protective factors in the case, while recognizing that those factors have not been validated for women (Cortoni, 2010; Hanson et al., 2026). The factors with some empirical support are a prior sexual offense and a noncontact sexual offense (Marshall et al., 2021); prior convictions, especially for offenses against children, including nonsexual child abuse (Sandler & Freeman, 2009; Cortoni, 2010); and solo offending against young male victims, a marker reported in an early Canadian study in which the only two sexual recidivists had offended alone (Cortoni, 2010; Hanson et al., 2026). In the New York sample, women with a prior drug, felony or misdemeanor conviction had several times the odds of sexual rearrest of women without one, but there were only 32 sexual recidivists in all (Sandler & Freeman, 2009). Hanson et al. (2026) stated that how much these factors raise risk is not known.

Cortoni (2010) described the clinical domains an assessment should address: sexual interest and the role of sex in the woman’s life; offense-supportive beliefs; relationships, including domestic violence and dependence on a partner; victimization history, which she cautioned should not be assumed to be the cause of offending; and general antisocial features. Whether a co-offender was involved, and whether the woman was coerced, a willing participant or the initiator, bears on treatment needs (Cortoni, 2010; see “Women Who Co-Offend With Male Partners”). A clinical study of 43 women referred to a specialist United Kingdom service after convictions or family-court findings described frequent histories of abuse, low self-esteem, cognitive distortions about children and relationship instability, and organized them as candidate static, dynamic and protective factors (Elliott et al., 2010). The study had no recidivism outcome, so its factors are hypotheses rather than validated predictors. Cortoni (2010) also cautioned that “higher risk” for a woman means higher than other women who sexually offend, not higher than men.

Newer instruments

Two instruments have been reported to predict sexual recidivism in samples that included women: the Iowa Sex Offender Risk Assessment, in a study with two female sexual recidivists among 105 women, and the Adult Dynamic Validated Instrument for Sex Offense Recidivism (ADVISOR), in a multistate sample with 14 female sexual recidivists among 442 women (Duwe et al., 2024; Hanson et al., 2026). Hanson et al. (2026) suggested that prudent evaluators may wish to wait for further research before using the ADVISOR with women.

Limits of the Evidence

All of these studies measure detected offending. Women made up about 2.2% of sexual offenders in police and court records and, on average, about 11.6% of perpetrators in victimization surveys that varied widely (about 3% to 24%) (Cortoni et al., 2017), which suggests that offenses by women are less often reported, and some offenses by women already known to the system may not lead to charges. The same undercount affects studies of men, and whether it is larger for women is not known (Hanson et al., 2026). The observed rates are therefore minimum estimates for both sexes. The women studied were mostly U.S. samples released in the 1980s through the 2000s, mostly White, and mostly identified through official records, often with follow-up overlapping the recruitment period. Several studies counted offenses involving the prostitution of minors as sexual offenses, which changes the rate. Comparisons with men usually draw on different samples, times and definitions, although the direction of the difference held in studies that measured both sexes the same way, with one exception. Because so few women reoffend sexually, the risk-factor findings rest on a few dozen recidivists across all studies, and no protective-factor research specific to women was located. Victim sex was examined mainly as a possible risk marker (solo offending against boys), and the research did not test whether rates differ by the sex of the victims more generally. Group rates cannot by themselves establish anything about an individual woman; that requires an individualized evaluation of her history, offense and circumstances.

Questions to Ask About This Research

When a figure such as “1% to 3%” is cited, counsel should ask what was counted (rearrest, charge or conviction), over what follow-up, and whether offenses involving the prostitution of minors were included; the New York rate was about 1.2% for women convicted of other sexual offenses and about 13% for the small group convicted of offenses involving the prostitution of minors (Cortoni, 2010). When the 17.5% rate from Nebraska (Bader et al., 2010) is cited, counsel should ask how large the sample was and how it compares with the meta-analytic average. When a Static-99R or STABLE-2007 score for a woman is offered, counsel should ask whether the instrument was designed for her: its rules exclude women, and a direct test in 739 women found no relationship between the total score and sexual recidivism (Marshall et al., 2021). When an LSI-R or similar score is offered, counsel should ask whether the conclusion follows: it bears on general reoffending, not specifically on sexual reoffending. When a woman is described as “high risk,” counsel should ask, high compared with whom. These questions are explained in “Reading Research Claims: A Guide for Attorneys” on this site.

Risk Assessment for Women and the Law

Federal and Military Standards

Expert risk testimony in federal court is governed by Federal Rule of Evidence 702, amended December 1, 2023, to make clear that the proponent must show by a preponderance of the evidence that the opinion rests on sufficient facts or data and reflects a reliable application of reliable methods, as applied in Daubert v. Merrell Dow Pharmaceuticals, Inc. (1993) and Kumho Tire Co. v. Carmichael (1999). For a woman, the threshold questions are whether any instrument used was validated for women and whether the evaluator’s method of adjusting a base rate is explained and grounded in the research. A score from an instrument whose own rules exclude women is unlikely to satisfy the reliable-application requirement on its own.

Federal civil commitment under 18 U.S.C. § 4248 requires proof by clear and convincing evidence that a person is “sexually dangerous to others,” defined in 18 U.S.C. § 4247(a)(6) as having a serious mental illness, abnormality or disorder as a result of which the person would have serious difficulty refraining from sexually violent conduct or child molestation; the Supreme Court upheld the statute in United States v. Comstock (2010). Kansas v. Hendricks (1997) and Kansas v. Crane (2002) require, for state commitment schemes, proof of a mental abnormality and of serious difficulty controlling behavior. Nothing in these statutes or decisions distinguishes women from men, but the evidentiary basis for predicting a woman’s sexual reoffending is much thinner. At federal sentencing, protection of the public is one of the purposes listed in 18 U.S.C. § 3553(a)(2)(C), and risk evidence is usually offered under that heading.

In courts-martial, Military Rule of Evidence 702 applies a parallel reliability standard; the 2023 federal amendment took effect in the military rule in 2025 (Joint Service Committee on Military Justice, 2025), and practitioners should confirm the current text. At presentencing, R.C.M. 1001(b)(5) allows trial counsel to offer opinion evidence of the accused’s rehabilitative potential, which must rest on sufficient knowledge of the accused’s personal circumstances and may not be based principally on the nature of the offense; the defense may offer evidence in extenuation and mitigation, which may include expert evidence about risk. Women are a small minority of accused in sexual offense courts-martial, and no research specific to military women who sexually offend was located.

Arizona

Arizona Rule of Evidence 702 was amended, effective January 1, 2024, to conform to the 2023 federal amendment (Order R-23-0004). Arizona’s sexually violent person statute applies to a person who has been convicted of, or found guilty except insane of, a sexually violent offense, or was charged with one and found incompetent, and who has a mental disorder that makes the person “likely to engage in acts of sexual violence” (A.R.S. § 36-3701). The statute defines a mental disorder as a paraphilia, personality disorder or conduct disorder, or a combination of them, that predisposes the person to commit sexual acts to such a degree as to render the person a danger to others. Like the federal statute, it does not distinguish by sex. Arizona registration requirements under A.R.S. § 13-3821 apply to women and men alike. Under A.R.S. § 13-3825, registrants are assigned community notification levels on the basis of a risk assessment; the statute does not name the instrument. Counsel in a case involving a woman should determine what instrument was used and whether it was developed or validated with women.

Special Issues in Legal Practice

The principles described above apply with equal force to every party in a legal proceeding. The science operates the same way regardless of who is asking the questions, and the research summarized here does not by itself favor either side. What differs between the parties is not the science but the vantage point — the particular places where these issues tend to surface, and the particular errors each party is best positioned to guard against. The two summaries below identify those recurring issues. They are offered for orientation, not as litigation strategy; in every case the controlling question is what the evidence in that case will, and will not, support, and the expert’s role is to inform the trier of fact rather than to advocate for either side.

Special Issues for Prosecutors

The research supports several points. Women do sexually reoffend; across studies 3.1% were detected doing so, the rates varied widely between samples, and one study that added child-welfare records found substantially more later sexual misconduct than charges alone revealed. Detected rates undercount offending, and offenses by women appear to be underreported. The factors with some support, including prior sexual offenses, noncontact offenses, prior convictions involving children and solo offending against young boys, are legitimate subjects of inquiry. About 30% of these women had some new criminal charge or conviction, which is relevant to supervision and to the protection of the public. Low recidivism figures do not diminish the harm of the index offense, which is assessed separately.

The research also counsels caution. Scoring a woman on the Static-99R or STABLE-2007, or citing male recidivism figures, is inconsistent with the instruments’ own rules and with the one direct test, and invites exclusion or impeachment under Rule 702. The risk factors identified for women rest on small numbers and have not been combined into a validated tool, so an evaluator who describes a woman as high risk should be able to explain the comparison group and the basis for the adjustment. A general-recidivism score should not be presented as a measure of sexual reoffending.

Special Issues for Defense

The research supports presenting the base rates accurately: in the most recent meta-analysis, 2.4% of women were detected committing a new sexual offense within five years, and the rate of new detections was lower among women who had remained offense-free for five years. It supports asking whether an opinion relies on a male instrument, male norms or male recidivism figures, and whether the evaluator considered gender-responsive needs such as trauma, mental health, substance use, relationships and support. Where a woman’s conviction involved a co-offender who coerced her, the research on co-offending bears on treatment needs and possibly on mitigation.

The research also counsels caution. Low group rates rest on detected offenses and cannot show that a particular woman poses no risk; the Nebraska findings show that official records can miss later conduct, and the variation across studies is large. Some markers of elevated risk, such as prior offenses against children or solo offending against young male victims, may be present in a given case. A low risk estimate does not reduce the seriousness of the offense or its effect on the victim, and an argument that treats it as doing so is not supported by the research.

This paper is for educational purposes only and is not legal advice. The author is a psychologist, not an attorney; summaries of law are general and must be verified against current authority or with a licensed attorney.

References

Association for the Treatment and Prevention of Sexual Abuse. (2025). Best practice guidelines for the assessment, treatment, risk management, and risk reduction of men who have committed sexually abusive behaviors (6th ed.). https://www.atsa.com

Bader, S. M., Welsh, R., & Scalora, M. J. (2010). Recidivism among female child molesters. Violence and Victims, 25(3), 349–362. https://doi.org/10.1891/0886-6708.25.3.349

Bouchard, K. N., Moulden, H. M., & Lalumière, M. L. (2019). Assessing paraphilic interests among women who sexually offend. Current Psychiatry Reports, 21(12), Article 121. https://doi.org/10.1007/s11920-019-1112-2

Cortoni, F. (2010). The assessment of female sexual offenders. In T. A. Gannon & F. Cortoni (Eds.), Female sexual offenders: Theory, assessment and treatment (pp. 87–100). Wiley-Blackwell. https://doi.org/10.1002/9780470666715.ch6

Cortoni, F., Babchishin, K. M., & Rat, C. (2017). The proportion of sexual offenders who are female is higher than thought: A meta-analysis. Criminal Justice and Behavior, 44(2), 145–162. https://doi.org/10.1177/0093854816658923

Cortoni, F., Hanson, R. K., & Coache, M.-È. (2010). The recidivism rates of female sexual offenders are low: A meta-analysis. Sexual Abuse, 22(4), 387–401. https://doi.org/10.1177/1079063210372142

Duwe, G., Kim, K., & Oglesby-Neal, A. (2024). The development and validation of the Adult Dynamic Validated Instrument for Sex Offense Recidivism (ADVISOR). Justice Evaluation Journal, 7(1), 35–58. https://doi.org/10.1080/24751979.2023.2200966

Elliott, I. A., Eldridge, H. J., Ashfield, S., & Beech, A. R. (2010). Exploring risk: Potential static, dynamic, protective and treatment factors in the clinical histories of female sex offenders. Journal of Family Violence, 25(6), 595–602. https://doi.org/10.1007/s10896-010-9322-8

Fortunato, E., Slikboer, R., Henshaw, M., Meyer, D., & Ogloff, J. R. P. (2026). Beyond the male lens: A comparison of females processed by police for online and offline child sexual abuse. Child Maltreatment. Advance online publication. https://doi.org/10.1177/10775595261453526

Hanson, R. K., Cortoni, F., & Sandler, J. (2026). The sexual recidivism rates of women are still low: An updated meta-analysis. Criminal Behaviour and Mental Health, 36(2), 53–62. https://doi.org/10.1002/cbm.70014

Joint Service Committee on Military Justice. (2025). Amendments to the Military Rules of Evidence since 1984 (G. E. Maggs, Ed.; 2025 ed.). https://jsc.defense.gov/Portals/99/MRE%20Amendments%202025.pdf

Marshall, E. A., & Miller, H. A. (2019). Examining gender-specific and gender-neutral risk factors in women who sexually offend. Criminal Justice and Behavior, 46(4), 511–527. https://doi.org/10.1177/0093854818796872

Marshall, E., Miller, H. A., Cortoni, F., & Helmus, L. M. (2021). The Static-99R is not valid for women: Predictive validity in 739 females who have sexually offended. Sexual Abuse, 33(6), 631–653. https://doi.org/10.1177/1079063220940303

Phenix, A., Fernandez, Y., Harris, A. J. R., Helmus, M., Hanson, R. K., & Thornton, D. (2017). Static-99R coding rules: Revised—2016 (Research Report 2017-R012). Public Safety Canada.

Sandler, J. C., & Freeman, N. J. (2009). Female sex offender recidivism: A large-scale empirical analysis. Sexual Abuse, 21(4), 455–473. https://doi.org/10.1177/1079063209347898

Saradjian, J. (2010). Understanding the prevalence of female-perpetrated sexual abuse and the impact of that abuse on victims. In T. A. Gannon & F. Cortoni (Eds.), Female sexual offenders: Theory, assessment and treatment (pp. 9–30). Wiley-Blackwell. https://doi.org/10.1002/9780470666715.ch2

Smith, P., Cullen, F. T., & Latessa, E. J. (2009). Can 14,737 women be wrong? A meta-analysis of the LSI-R and recidivism for female offenders. Criminology & Public Policy, 8(1), 183–208. https://doi.org/10.1111/j.1745-9133.2009.00551.x

Van Voorhis, P., Wright, E. M., Salisbury, E., & Bauman, A. (2010). Women’s risk factors and their contributions to existing risk/needs assessment: The current status of a gender-responsive supplement. Criminal Justice and Behavior, 37(3), 261–288. https://doi.org/10.1177/0093854809357442

Legal Authorities

Daubert v. Merrell Dow Pharmaceuticals, Inc., 509 U.S. 579 (1993).

Kansas v. Crane, 534 U.S. 407 (2002).

Kansas v. Hendricks, 521 U.S. 346 (1997).

Kumho Tire Co. v. Carmichael, 526 U.S. 137 (1999).

United States v. Comstock, 560 U.S. 126 (2010).

18 U.S.C. § 3553(a).

18 U.S.C. §§ 4247(a)(5)–(6), 4248.

Ariz. Rev. Stat. Ann. §§ 13-3821, 13-3825, 36-3701.

Fed. R. Evid. 702.

Ariz. R. Evid. 702.

Mil. R. Evid. 702.

Rule for Courts-Martial 1001.

Related topic papers: Women Who Sexually Offend: Prevalence and Underreporting · Typologies and Offense Pathways of Women Who Sexually Offend · Women Who Co-Offend With Male Partners · Risk Assessment Tools Outside Their Normative Samples · Gender and the Justice System’s Response to Women Who Sexually Offend

Related topics from Arizona Forensic Services: Psychosexual Risk Evaluations · Online Sex Offense · Courts-Martial