The study hypothesized that the sociodemographic and criminogenic factors available in most court and probation records, in conjunction with formal and informal social controls, could predict nonsexual recidivism among male sex offenders on probation. The second hypothesis was that this same combination of factors would not result in a statistically significant prediction tool for identifying those sex offenders most likely to be rearrested for a new sex crime while on probation. Data for this analysis were collected by the U.S. Bureau of Justice Statistics between 1986 and 1989. The follow-up study conducted in 1987, 1988, and 1989 was used for 32 counties in 17 States randomly selected from the original 100 counties; it involved 81,927 probationers out of the 306,000 probationers sentenced in the year the study began. Sentencing records, probation files, and criminal history records were the primary information sources. Self-report data through probationer questionnaires were also used. The overall recidivism rate was 16 percent; however, only 4.5 percent of the offenders committed a new sex crime during probation. Regression analyses indicate that factors readily available to court personnel can accurately predict recidivism for nonsexual offenses by sex offenders on probation. Accurately predicting new violent sex crimes, however, was not successful, thus supporting one hypothesis. Although probation apparently is suitable for deterring new sex crimes by a significant number of sex offenders, it is important to develop an accurate tool for predicting the likelihood of new sex offenses. This will require further study on the precursors to various types of sex crimes during a probation term, as well as the factors associated with their deterrence. 3 tables, 8 notes, and 44 references