The relationship between serum testosterone and sexual violence was examined in a sample of 501 convicted adult male sex offenders attending an intensive in-hospital group psychotherapy treatment program. It was found that men with higher testosterone tended to have committed the most invasive sexual crimes (p < .001, two-tailed). Further, a positive partial correlation (controlling for age) between testosterone and sexual offense recidivism over a lengthy follow-up period (mean = 8.9 years) was found. When the sample was separated into one group that completed treatment and one group that did not, an important ameliorating treatment effect was observed. Although controlling for age, serum testosterone remained significantly predictive of sexual recidivism for the treatment noncompleter group (p < .05, two-tailed). For those who completed treatment testosterone was no longer predictive of sexual reoffense (p > .05, two-tailed). Among convicted sex offenders, higher serum testosterone appears to be associated with greater likelihood of further sexual violence. Effective therapy, however, appears able to intercede in the influence of testosterone on sexually deviant behavior. It is suggested that serum testosterone may be an informative static risk factor and completion of intensive treatment should be accorded significance in future actuarially based risk prediction instruments.