The 10 subsamples of sex offenders ranged in size from 191 to 1,138 and were drawn from the Canadian jurisdictions of Quebec, Ontario, Manitoba, Alberta, and the Correctional Service of Canada, as well as the U.S. States of California and Washington State, and Great Britain's Her Majesty's Prison Service (England and Wales). All of the offenders in the subsamples had been released from correctional institutions, with the exception of the Manitoba probation sample and about half of the offenders from the Washington sample who received community sentences. Case specific information, without individual identifiers, from the original 10 datasets were merged for the current analysis. Recidivism estimates were computed by using survival analysis. This analysis produced the cumulative proportion surviving without offenses at the end of a specific time period. These survival percentages were then subtracted from 100 to produce estimates of the recidivism potential at 5-, 10-, and 15-year intervals. "Sexual recidivism" was measured by using the original definitions from the research reports: five datasets used convictions; four datasets used new charges or a new conviction; and one sample used convictions, charges, and additional police information. The 5- and 10-year recidivism estimates were 17 percent and 21 percent for the studies that used only convictions as their recidivism criteria, and 12 percent and 19 percent for the studies that used charges and convictions as their recidivism criteria. Given the similarity in the recidivism rates based on convictions alone and charges and convictions, the data were combined to provide overall estimates of sexual recidivism rates. The rates estimated using the combined sample would be closer to the reconviction rate than the rearrest rate, because apparently the sources used for the recidivism data contained relatively few charges that did not ultimately result in conviction. The findings indicate that most sex offenders did not repeat sex offenses, that first-time sex offenders were significantly less likely to repeat sex offenses than those with previous convictions for sex offenses, and that offenders over the age of 50 were less likely to reoffend than younger offenders. In addition, the study found that the longer offenders remained offense-free in the community, the less likely they were to reoffend sexually. The data show that rapists, incest offenders, "girl-victim" child molesters, and "boy-victim" child molesters recidivated at significantly different rates. Recidivism rates were highest for "boy-victim" child molesters and rapists. Generally, the sex offenders that posed the highest risk of recidivism victimized males, had prior sex offenses, and were a younger age. Given that the level of sexual recidivism was lower than commonly believed, discussions of the risk posed by sex offenders should differentiate between the high public concern about these offenses and the relatively low probability of sexual recidivism. Further, the variation in recidivism rates suggests that not all sex offenders should be treated the same. 3 tables, 6 figures, and 37 references