One treatment program is delivered in specialized living units for JSOs that separates them from the general juvenile inmate population. Participants in the second treatment program remain housed with the general population of juvenile inmates. Arrest and incarceration data through January 2003 were obtained for 261 male JSOs released between 1992 and 2001. The inclusion of incarceration data for the sample during adulthood allowed for a more accurate assessment of the actual time at risk for sexual reoffending. Outcome data included rearrest rates, length of time to rearrest, and type of offense (property, nonsexual assault, sexual) for which rearrests occurred. Analyses used survival curve functions. For both treatment groups, rearrest was most likely to involve a nonsexual person offense (31 percent and 47 percent, respectively) and least likely to involve a sexual offense (less than 5 percent for both groups). The group that received treatment in a self-contained housing unit had a lower rearrest rate and a longer mean time to rearrest for all types of offenses than the treatment group housed with the general inmate population. Also, juveniles assessed to have high levels of impulsive/antisocial behaviors were significantly more likely to reoffend compared to juveniles with low levels of such behaviors, regardless of treatment type. 4 tables, 2 figures, and 36 references