This study challenges a number of preconceptions about the risks posed by sex offenders who have been sentenced to long determinate terms of imprisonment: 162 prisoners were followed-up for four years and 94 for six years. The category `sexual offender¿ was disaggregated in order to examine reconviction rates of offenders against adults as compared with offenders against children, whether in an intra-familial or extra-familial setting, and to explore evidence of `specialization¿. The study also analysed the extent to which members of the Parole Board, in deliberating on the suitability of these prisoners for parole, correctly identified as `high risks¿ those who were subsequently reconvicted of a sexual or serious violent crime. These `clinical¿ predictions were compared with those derived from an actuarial prediction instrument for sex offenders. The findings have implications for both sentencing and parole.