A total of 178 sex offenders who had been assessed at a maximum-security psychiatric facility were followed for an average of 59 months of opportunity to reoffend. Of these subjects, 124 were defined as child molesters, including 13 exclusively incestuous offenders, 28 as rapists, and 26 as both. Data were obtained on demographic variables, psychiatric records, criminal history, sexual preference, and outcome variables. Recidivism was defined as a sexual conviction or as a violent failure. Findings show that 28 percent of the subjects were convicted of a new sex offense, and 40 percent were arrested, convicted, or returned to the psychiatric facility for a violent, including sex, offense. Rapists were more likely to recidivate than child molesters. Psychopathy, measures of previous criminal history, and phallometric indexes of deviant sexual interests were found to be useful predictors of sexual recidivism. A predictor scale was developed by weighting each predictor that was significant in the regression analyses by a number that reflected its univariate correlation with the criterion. A linear relationship (r=.45) was found between scores on the predictor scale and reconviction for a sexual offense. 4 tables, 3 figures, and 34 references