This paper reviews some of the literature on the prediction of violent or sexual recidivism among released sex offenders. Several prediction tools are examined. They are: Statistical Index of General Recidivism Scale, The Level of Supervision Inventory, The Risk Checklist for Child Molesters, The Psychopathy Checklist-Revised, and The Statistical Prediction of Violent Recidivism Instrument. It is concluded that it may be possible to obtain reasonably accurate predictions of sexual or violent re-offence among released violent sexual offenders using The Psychopathy Checklist-Revised and The Statistical Prediction of Violent Recidivism Instrument. However, the applicability of these instruments to specific populations of sex offenders is yet to be demonstrated. Survival curves or expectancy tables are needed to adjust the predictions for shorter periods of time than are considered in the follow-up studies on which the tools have been based, and to resolve apparent discrepancies across studies. General observations are made about the factors which are predictive of re-offence when the offence type of the offender is controlled. Cautions are raised about the underestimation by acturial prediction of the true rate of re-offence and of the degree of harm which may be inflicted by sexual sadists or sexual murderers.