In late 1993 Oklahoma Department of Corrections officials implemented a cognitive behavioral treatment program through the correctional system. Earlier evaluations of MRT used in a Tennessee therapeutic community yielded promising results. Relying on official records of institutional misconduct and community recidivism, this analysis of Oklahoma's implementation of MRT compared the behavior of individuals who participated in MRT with individuals who participated in other programs, as well as with individuals who did not participate in any programming. The longitudinal nature of the analysis also facilitated an examination of behavioral changes within individuals before and after they began programs. Results of the analysis show that the decision to participate in MRT and in other programs was the outcome of a nonrandom process; closure on the issue of group comparisons is premature, because treatment effects and factors that lead to initiation are difficult to disentangle; and among individuals who participated in MRT, the program apparently yielded moderate but statistically significant reductions in the risk of misconduct and recidivism. Limitations of the analysis include a need to better understand the process that leads to program initiation, large quantities of missing data, and relatively short follow-up periods (less than 1 year in most cases). Extensive tabular data and figures and 152 references