Sexual deviance measured phallometrically can discriminate with high accuracy between sex offenders and other men. Sex offenders remain at risk for a long time after release; researchers have developed combinations of variables selected for the cumulative efficiency in predicting recidivism. Actuarial assessments offer the most empirically valid way to classify offenders according to risk. However, actuarial instruments and other methods cannot identify all dangerous sex offenders without an unacceptably high false-positive rate. In addition, the ability of any interventions to reduce recidivism is unknown. Existing data do not support the opinion that treatment reduces risk; volunteering for treatment and persisting with it are indicators of lower risk. Almost nothing is known about the effects of intensive supervision of sex offenders, registration, or community notification. Nevertheless, offenders should be offered clinically defensible intervention strategies, and new treatment strategies should be developed and evaluated. In addition, methodologically sound evaluations of all interventions (therapy, drugs, community notification, registration, and supervision) should receive high priority. Finally, application of the available data can increase the likelihood that new legislation intended to reduce the risk represented by sexual predators will have its intended results. Figures, footnote, and 163 references