The present study examined whether there were program differences with respect to post-release outcomes in 20 federal in-prison substance abuse programs which used a cognitive-behavioral treatment approach. Recidivism and post-release drug use were examined for a sample of 1,343 individuals--1,065 men and 278 women. Discrete time proportional hazards regression models showed that, after controlling for individual characteristics, no differences were detected among the 16 programs for men. In contrast, one of the four female programs had significantly higher drug use rates and one had significantly lower recidivism rates. Our results suggest that implementation of a treatment approach which has been shown to be effective--cognitive-behavioral treatment--can result in comparable outcomes across programs, despite differences in program implementation. However, specific types of variation in program implementation may affect outcomes. (Contains 3 tables and 5 notes.)