This study used four Cox regression analyses to examine the predictors of reoffending in a sample of 86 male adolescents adjudicated delinquent for the commission of sexual offenses and treated in a corrections-based sex offender treatment program. The results indicate that the predictors of reoffense, when defined as an arrest, conviction, or parole violation for any new crime, differ somewhat from those found for adults. Specifically, increased risk for reoffense was associated with impulsivity, involvement with significantly younger children, younger age at first offense, and shorter treatment stays. Decreased risk for reoffense was associated with having a male victim, having been a sexual abuse victim, and multiple paraphilias. These factors were not all stable across analyses. These data indicate that the risk prediction methods used for adult sex offenders would not be appropriate for adolescent populations and that more research is needed before attempts are made to develop such tools for adolescents.