Results indicate that theories on deviance and treatment motivation are useful not only to predict treatment outcomes as they have been traditionally used, but also to understand the paths of engagement that lead to those outcomes. By testing five hypotheses, 58 percent of the few compliant clients were identified, as well as 88 and 55 percent of low- and high-rate noncompliant clients, respectively. This study sought to build on recent research on the treatment process by summarizing five hypotheses of treatment noncompliance: physical prime, supportive social network, conventional social involvement, treatment motivation, and risk-taking propensity. Testing them through the analysis of treatment data from 150 criminal justice-mandated clients of long-term therapeutic communities. The hypotheses most useful in explaining variations in treatment compliance were the physical prime and supportive social networks. Conventional social involvement and treatment motivation hypotheses were also partially validated by the data. Client age was the strongest and most consistent individual correlate of treatment compliance, with the specific dynamics of this relationship worthy of more study. Compliance with therapeutic regimes constitutes an important but infrequently studied precursor of treatment engagement and a necessary condition of successful treatment. Lessons learned mainly apply to repeat drug-involved criminal offenders, who have been the principal driving force behind the growth of the American prison population in the last two decades. Tables, references