Snap Shot
- The question: How often do adolescents who have sexually offended commit another sexual offense, how should those rates be defined and read, and what do they say about the smaller group whose offending persists?
- What the research shows: In meta-analyses of official records, detected sexual recidivism averaged about 7% over roughly five years in 63 data sets (Caldwell, 2010) and 4.92% in 106 data sets over about five years, with a lower average (2.75%) in studies conducted from 2000 to 2015 (Caldwell, 2016); individual recent samples report higher detected rates (for example, 9.5% to 10.8% in a Queensland evaluation and 16% at about 16 years in a Canadian sample), rates were higher with longer follow-up in some samples, and detected recidivism undercounts offending. Nonsexual reoffending is several times more common than sexual reoffending, most adolescents who sexually offend are not later identified as adult sex offenders, and a small subgroup persists.
- What it does not show: Official records miss offenses that are never reported or charged, so detected rates understate actual reoffending by an unknown amount. Low group rates do not show that a particular adolescent will not reoffend, and rates from one era, country, or setting may not fit another.
- The law: Recidivism evidence bears on federal and Arizona transfer decisions (18 U.S.C. § 5032; A.R.S. §§ 8-327, 13-504), on registration of juveniles (34 U.S.C. § 20911(8); A.R.S. § 13-3821), on federal sentencing under 18 U.S.C. § 3553(a), and, in courts-martial, on rehabilitative-potential evidence, which R.C.M. 1001 requires to rest on the accused’s personal circumstances.
- Issues for each side: Prosecutors should not present outdated or adult recidivism figures as the risk of an adolescent, but may properly point to undetected offending and to the features of the persistent subgroup; defense counsel can rely on low and declining detected rates while recognizing that they are minimum estimates and do not describe every adolescent.
Whether an adolescent who has sexually offended is likely to do so again is a central question in juvenile disposition, transfer to adult court, registration, treatment planning, and the sentencing of young adults whose offenses began in adolescence. For many years policy assumed that adolescents who sexually offend were adult sex offenders in the making. Retrospective studies of adults in treatment, many of whom reported that their offending began in adolescence, contributed to that belief (Lobanov-Rostovsky, 2015), and prospective research has since produced a different picture.
This paper explains how sexual recidivism is defined and measured in this research, summarizes the meta-analytic and long-term follow-up data, describes the decline in recidivism rates over recent decades, and addresses two points that cut in opposite directions: detected recidivism undercounts reoffending, and most adolescents who sexually offend are not later identified as adult sex offenders. Risk assessment instruments are addressed in “Assessing Risk in Adolescents: ERASOR, J-SOAP-II, and PROFESOR,” and the characteristics of these adolescents in “Normative and Problematic Sexual Behavior in Children and Adolescents.”
What a Recidivism Rate Measures
A recidivism rate is the percentage of a defined sample who, within a defined follow-up period, are recorded as committing a new offense of a defined kind. Each element affects the number. The outcome may be a new arrest, charge, conviction, or return to custody; may be limited to sexual offenses or include any offense; and may be drawn from juvenile records, adult records, or both. Follow-up may begin at adjudication, release, or the eighteenth birthday, and may last one year or twenty. Samples may come from community supervision, residential treatment, secure custody, or a whole birth cohort, and from different countries and decades. A rate stated without these details cannot be compared with another rate or applied to a particular adolescent.
All of these studies measure detected offending. Official records miss offenses that victims do not report, that police do not clear, and that prosecutors do not charge, so observed rates are underestimates of actual reoffending (Lobanov-Rostovsky, 2015). That limitation applies to every study discussed below. It also applies in the other direction to a common misreading: a person who appears in a recidivism count has been recorded as reoffending under that study’s definition, which may include charges that did not result in conviction.
Meta-Analytic Base Rates
Caldwell (2010) analyzed 63 data sets totaling 11,219 adolescents who had sexually offended, with an average age of 14.8, followed for a weighted mean of 59.4 months through official records of arrest or conviction. The weighted mean sexual recidivism rate was 7.08% (range across studies, 0% to 18%), and the weighted mean general recidivism rate was 43.4%. Neither the setting from which the sample was drawn (community, residential, or secure custody) nor the use of arrest rather than conviction as the outcome significantly affected sexual recidivism rates. Studies that followed adolescents during adolescence found monthly sexual recidivism rates more than four times higher than studies relying only on adult records, so studies that begin follow-up at 18 miss much of the reoffending that occurs during adolescence.
A later meta-analysis by the same author examined 106 data sets with 33,783 adolescents adjudicated for sexual offenses in studies conducted between 1938 and 2014 (Caldwell, 2016). Across all studies, the weighted mean detected sexual recidivism rate was 4.92% over a mean follow-up of about five years, and general recidivism was about 39%. Longer follow-up yielded higher sexual recidivism rates up to about 36 months, with no significant further increase beyond that point. The author estimated the base rate across the full data set at between 3% and 10%, with a global average of about 5%.
A federal research brief prepared for the Department of Justice’s SMART Office summarized the field in 2015: observed sexual recidivism ranged from about 7% to 13% over about 59 months depending on the study, sexual recidivism was lower than general recidivism, and adolescents who sexually offend should not be presumed to be on a path to lifelong sexual offending (Lobanov-Rostovsky, 2015). The brief noted no clear difference in sexual or general recidivism between adolescents whose victims were younger children and those whose victims were peers or adults, while cautioning that the underlying studies varied in measurement, sample size, and follow-up.
The Decline in Recidivism Rates
Rates have fallen. Among studies that began follow-up after 1980, those conducted from 2000 to 2015 produced a weighted average of 2.75%, against 10.30% in studies from 1980 to 1995, a decline of about 73% (Caldwell, 2016). In Caldwell’s analysis, the decline remained after controlling for follow-up time and for the setting from which samples were drawn; the 2.75% figure comes from a subset of the 106 data sets (the number of data sets and youth in the 2000 to 2015 group is not reported in the abstract), and both of Caldwell’s meta-analyses were written by the same author. Older studies sometimes counted conduct, such as noncoercive sexual activity between similar-age adolescents, that would not be treated as a sexual offense today, and Caldwell excluded studies that began before 1980 from the trend analysis for that reason. He considered whether recent cohorts had been diluted with lower-risk adolescents, or whether fewer victims were reporting, and concluded from juvenile arrest rates, victimization surveys, and child maltreatment data that the decline likely reflected a real decline in sexually abusive behavior; these are indirect indicators, and changes in reporting and charging practices cannot be ruled out.
The decline is not confined to adolescents. In a meta-analysis of 468 studies from the United States and Canada published between 1940 and 2019, sexual recidivism rates fell by about 45% to 60% since the 1970s after accounting for study setting, follow-up length, and how recidivism was defined (Lussier & McCuish, 2024). About 16% of the studies came from juvenile justice settings; the drop remained when youth samples were removed, and rates from juvenile justice settings did not differ significantly from those in adult corrections.
Recent rates are not uniformly below 3%, however. In a Queensland, Australia, evaluation of adolescents found guilty of sexual offenses committed at ages 12 to 17 between 2001 and 2012, sexual recidivism, defined as offenses resulting in police action, was 9.5% among 200 males who completed a specialized treatment program and 10.8% among 295 who were not accepted or not referred, over average follow-ups of about five years for the treated group and six and a half for the comparison group; both groups had high rates of nonsexual reoffending (Ogilvie et al., 2024). The difference was not statistically significant, and the groups were not randomly assigned. Rates vary across jurisdictions and definitions, and a figure from one country or study should not be applied to another without examining how it was produced. A Wisconsin sample of 172 youth released from secure juvenile corrections had 12.2% charged with a new felony sexual offense over a mean of about four years (Caldwell & Dickinson, 2009), a higher-risk setting than most adjudicated adolescents come from.
Long-Term Follow-Up and the Persistent Subgroup
Longer studies show how the rate accumulates and who reoffends. In a Canadian program evaluation, 148 adolescents aged 12 to 19 (139 male, 9 female) were followed through a national database of criminal charges. Over a mean of 6.2 years, sexual recidivism was 5.17% among 58 who completed at least 12 months of specialized treatment and 17.8% among 90 who received only an assessment, refused treatment, or dropped out; sexual recidivism was predicted by sexual interest in children, while nonsexual recidivism was predicted by factors associated with general delinquency (Worling & Curwen, 2000). At an extended mean follow-up of 16.2 years, the treated adolescents remained significantly less likely to be charged with sexual offenses (9% of 58 treated vs. 21% of 90 comparison adolescents), nonsexual violent offenses, and nonviolent offenses (Worling et al., 2010), and the overall sexual recidivism rate for the full sample was 16% (Lobanov-Rostovsky, 2015; Worling et al., 2010). The comparison group was not randomized and included adolescents who refused or left treatment, which may have differed from those who completed it.
Birth-cohort studies avoid the selection that shapes clinical samples. In the 1984 Dutch birth cohort, police records for 87,528 men were examined from age 12 to 23 (Lussier & Blokland, 2014). Of the 341 registered for a sex crime in adolescence, 3.0% of those registered once and 12.3% of those registered more than once were registered for a sex crime in early adulthood. Conversely, only 17 of the 377 men registered for a sex crime in adulthood (4.5%) had been registered for one as adolescents, and men with no adolescent sex crime registration accounted for 91.3% of adult sex crimes. The frequency of general, nonsexual offending in adolescence was associated with a greater likelihood of adult sex offending. The authors concluded that most adolescents who sexually offend desist, that most adult sex offending begins in adulthood, and that a small group of adolescents with repeated sexual offenses is at risk of persisting. Earlier U.S. cohort research reached similar conclusions: in three Racine, Wisconsin, birth cohorts, 8.5% of males with a juvenile sex-offense police contact had an adult sex-offense contact, compared with 6.2% of males with only nonsexual juvenile contacts, and the frequency of juvenile offending predicted adult sex offending better than whether the juvenile offense was sexual (Zimring et al., 2007). Those authors drew policy conclusions skeptical of juvenile registration; the data themselves are police contacts for males only. The Dutch study included only men and followed them only to 23, so it cannot describe offending in later adulthood.
Undetected Offending
Every figure above counts only what reached official records, and evidence from self-report shows that the gap can be substantial. In a nationally representative U.S. sample of youth followed in the National Youth Survey, 58% of those who reported committing a sexual assault reported committing another, and 10% of those who reported a sexual assault as juveniles reported an adult sexual offense (Grotpeter & Elliott, 2002, as reported secondhand in Lobanov-Rostovsky, 2015; the original report was not reviewed for this paper, so the sample size and the survey’s definition of sexual assault are not stated here, and the respondents were a general-population sample, not adjudicated adolescents). In one study of adolescents who had sexually offended, reconviction was 6% but rose to 11% when self-reported reoffending was included (Bremer, 1992, as cited in Lussier & Blokland, 2014). Polygraph examinations in treatment have elicited disclosure of additional victims and offenses among adolescents, but disclosures obtained this way have not been independently validated, and adolescents are more prone than adults to false admissions under pressure (Caldwell, 2020).
These findings support the conclusion that official rates are minimum estimates. They do not establish a multiplier that converts a detected rate into a true rate, and they do not overturn the central finding of the prospective studies. As Lussier and Blokland (2014) observed, reliance on official data underestimates recidivism, but it cannot account for the fact that the great majority of adolescents who sexually offend are not rearrested for a sexual offense.
Limits of the Evidence
Most samples are North American or European, most are entirely or predominantly male, and many draw on studies that are decades old; female adolescents appear in small numbers, and their rates cannot be estimated separately from this literature. Studies differ in outcome definitions, follow-up periods, and settings, and meta-analytic averages conceal that variation. Most treatment comparisons are not randomized. Detected rates understate reoffending by an unknown amount, while self-report studies are few, small, or old. Group rates describe groups: they cannot establish whether a particular adolescent will or will not reoffend, which depends on individual assessment of factors such as sexual interest, persistence of sexual offending, general antisocial behavior, and supervision and support.
Questions to Ask About This Research
When the 2.75% figure (Caldwell, 2016) is offered, ask what was counted and over what period: detected arrests or convictions, averaged across studies conducted from 2000 to 2015 with follow-ups of about five years. When the 58% self-reported repeat figure from the National Youth Survey is offered, ask who was studied: a general-population sample reporting any sexual assault, not adolescents adjudicated and treated for sexual offenses. When an older rate in the 10% to 20% range is cited, ask whether the finding applies today, given the documented decline. When a treated-versus-untreated difference is cited (Worling & Curwen, 2000; Worling et al., 2010), ask whether the conclusion follows from the design, because the comparison group included adolescents who refused or left treatment. When Dutch or Racine cohort findings are cited, ask how long the follow-up lasted and whether women were included. These questions are explained in “Reading Research Claims: A Guide for Attorneys” on this site.
Adolescent Recidivism Research and the Law
Federal and Military Standards
A federal juvenile may be transferred for adult prosecution only on statutory grounds and after the court considers, among other factors, the juvenile’s age and social background, the nature of the offense, prior record, present intellectual development and psychological maturity, past treatment efforts, and available programs (18 U.S.C. § 5032). Under SORNA, a juvenile adjudication counts as a conviction requiring registration only if the juvenile was at least 14 and the offense was comparable to or more severe than aggravated sexual abuse (34 U.S.C. § 20911(8)). At adult sentencing, the need to protect the public from further crimes is one of the purposes in 18 U.S.C. § 3553(a)(2)(C), and expert testimony about risk must meet Federal Rule of Evidence 702, amended December 1, 2023, under Daubert v. Merrell Dow Pharmaceuticals, Inc. (1993) and Kumho Tire Co. v. Carmichael (1999). An opinion that applies a meta-analytic base rate to a defendant without examining whether the underlying samples resemble the defendant may be challenged under the amended rule.
The military has no juvenile system, but young service members may have juvenile histories or have offended in late adolescence. In sentencing, the prosecution’s opinion evidence on rehabilitative potential must be based on information about the accused and relate to the accused’s personal circumstances; the severity or nature of the offense may not be its principal basis, and the witness may not opine on whether a punitive discharge is appropriate (R.C.M. 1001(b)(5)). An opinion resting on group recidivism rates alone would be difficult to reconcile with that requirement. The defense presents matters in extenuation and mitigation under R.C.M. 1001(d), where recidivism research may also be offered. The 2023 federal amendment to Rule 702 took effect in Military Rule of Evidence 702 in 2025 (Joint Service Committee on Military Justice, 2025); practitioners should confirm current text. The R.C.M. provisions described here are from the 2024 Manual for Courts-Martial.
Arizona
When the State seeks transfer of a juvenile to adult court, the juvenile court decides by a preponderance of the evidence whether public safety would best be served by transfer, considering factors that include the seriousness of the offense, the juvenile’s record and previous history, the juvenile’s mental and emotional condition, and the likelihood of rehabilitation through services available to the juvenile court (A.R.S. § 8-327(C), (D)). A juvenile prosecuted as an adult under A.R.S. § 13-501(B) may seek transfer to juvenile court, which requires a finding by clear and convincing evidence that public safety and rehabilitation would best be served (A.R.S. § 13-504(A), (C)). Registration for a juvenile adjudicated delinquent for a listed offense is discretionary, ends at 25, and may be terminated on successful completion of probation (A.R.S. § 13-3821(D), (F), (G)). Recidivism research bears on each of these determinations, but only as background to an individual assessment.
A 2019 report to the Juvenile Justice Services Division of the Arizona Supreme Court, prepared by the National Center for State Courts and the Center for Sex Offender Management, recommended that disposition and case management be driven by assessment rather than by the offense alone (as described in Caldwell, 2020, an unpublished review that has not been peer reviewed). Arizona Rule of Evidence 702 was amended, effective January 1, 2024, to conform to the 2023 federal amendment (Order R-23-0004).
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 the position that detected rates are minimum estimates, that self-report studies show reoffending that never reaches official records, and that a subgroup of adolescents persists. Repeated sexual offending in adolescence, sexual interest in children, and extensive general offending are associated with later sexual offending in the studies reviewed here, and some recent samples, such as the Australian study, show rates near 10%. Prosecutors are often best positioned to document the number and pattern of offenses, any victims who disclosed late, offending during prior supervision, and the adolescent’s general offending, which bear on whether the adolescent resembles the persistent group.
The research also counsels caution. Older juvenile rates, adult recidivism rates, and retrospective accounts from adult offenders do not describe adolescents adjudicated today, and the self-report figures come from general-population samples rather than from adjudicated and treated adolescents. Testimony that converts an undetected-offending argument into a specific likelihood for a particular adolescent goes beyond the data, and in courts-martial R.C.M. 1001(b)(5) bars a rehabilitative-potential opinion that rests principally on the nature of the offense.
Special Issues for Defense
The research supports the position that most adolescents who have sexually offended are not detected committing another sexual offense, that detected rates have declined substantially, that nonsexual reoffending is more common than sexual reoffending, and that, in the Dutch and Racine cohorts, most adult sex offending was committed by men with no adolescent sex offense record. Defense counsel are often best positioned to gather treatment records, school and family history, and evidence of desistance, and to ask whether a risk opinion considered the decline in rates and the developmental differences between adolescents and adults.
The research also counsels caution. Detected rates understate reoffending, the self-report evidence suggests the gap can be meaningful, and favorable averages do not show that a particular client is outside the persistent subgroup. Repeated sexual offending, sexual interest in children, or extensive general delinquency in the record will not be offset by citing a group base rate, and treatment-outcome comparisons drawn from nonrandomized samples do not show that treatment alone will prevent reoffending in a particular case.
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
Caldwell, M. F. (2010). Study characteristics and recidivism base rates in juvenile sex offender recidivism. International Journal of Offender Therapy and Comparative Criminology, 54(2), 197–212. https://doi.org/10.1177/0306624X08330016
Caldwell, M. F. (2016). Quantifying the decline in juvenile sexual recidivism rates. Psychology, Public Policy, and Law, 22(4), 414–426. https://doi.org/10.1037/law0000094
Caldwell, M. F. (2020, March 4). Review of the final report: Enhancing court and probation practices for cases involving juveniles who commit sexually abusive behaviors [Unpublished report]. Department of Psychology, University of Wisconsin–Madison.
Caldwell, M. F., & Dickinson, C. (2009). Sex offender registration and recidivism risk in juvenile sexual offenders. Behavioral Sciences & the Law, 27(6), 941–956. https://doi.org/10.1002/bsl.907
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
Lobanov-Rostovsky, C. (2015). Recidivism of juveniles who commit sexual offenses (SOMAPI Research Brief). U.S. Department of Justice, Office of Justice Programs, Office of Sex Offender Sentencing, Monitoring, Apprehending, Registering, and Tracking.
Lussier, P., & Blokland, A. (2014). The adolescence-adulthood transition and Robins’s continuity paradox: Criminal career patterns of juvenile and adult sex offenders in a prospective longitudinal birth cohort study. Journal of Criminal Justice, 42(2), 153–163. https://doi.org/10.1016/j.jcrimjus.2013.07.004
Lussier, P., & McCuish, E. (2024). Revisiting the sexual recidivism drop in Canada and the United States: A meta-analysis of 468 empirical studies involving 388,994 individuals. Journal of Criminal Justice, 92, Article 102188. https://doi.org/10.1016/j.jcrimjus.2024.102188
Ogilvie, J. M., McKillop, N., Cale, J., Allard, T., Rynne, J., & Smallbone, S. (2024). Assessing the effectiveness of a specialized, field-based treatment program for youth who have committed sexual offenses in an Australian jurisdiction. International Journal of Offender Therapy and Comparative Criminology, 68(15), 1540–1557. https://doi.org/10.1177/0306624X231219216
Worling, J. R., & Curwen, T. (2000). Adolescent sexual offender recidivism: Success of specialized treatment and implications for risk prediction. Child Abuse & Neglect, 24(7), 965–982. https://doi.org/10.1016/S0145-2134(00)00147-2
Worling, J. R., Litteljohn, A., & Bookalam, D. (2010). 20-year prospective follow-up study of specialized treatment for adolescents who offended sexually. Behavioral Sciences & the Law, 28(1), 46–57. https://doi.org/10.1002/bsl.912
Zimring, F. E., Piquero, A. R., & Jennings, W. G. (2007). Sexual delinquency in Racine: Does early sex offending predict later sex offending in youth and young adulthood? Criminology & Public Policy, 6(3), 507–534. https://doi.org/10.1111/j.1745-9133.2007.00451.x
Legal Authorities
Daubert v. Merrell Dow Pharmaceuticals, Inc., 509 U.S. 579 (1993).
Kumho Tire Co. v. Carmichael, 526 U.S. 137 (1999).
18 U.S.C. §§ 3553(a), 5032.
34 U.S.C. § 20911(8).
Ariz. Rev. Stat. Ann. §§ 8-327, 13-501(B), 13-504, 13-3821(D), (F), (G).
Fed. R. Evid. 702.
Mil. R. Evid. 702.
Ariz. R. Evid. 702.
Rule for Courts-Martial 1001(b)(5), (d).
Related topic papers: Normative and Problematic Sexual Behavior in Children and Adolescents · Assessing Risk in Adolescents: ERASOR, J-SOAP-II, and PROFESOR · Juvenile Transfer and Competency in Sex Offense Cases · Juvenile Registration and the Research on Its Effects · Young Adults Aged 18 to 25 · Treatment for Adolescents Who Sexually Offend
Related topics from Arizona Forensic Services: Psychosexual Risk Evaluations · Courts-Martial