The present report is an intent-to-treat analysis involving secondary data drawn from the first randomized clinical trial of prison-initiated methadone in the United States. This study examined predictors of treatment entry and completion in prison. A sample of 211 adult male prerelease inmates with preincarceration heroin dependence were randomly assigned to one of three treatment conditions: counseling only (counseling in prison; n = 70); counseling plus transfer (counseling in prison with transfer to methadone maintenance treatment upon release; n = 70); and counseling plus methadone (methadone maintenance in prison, continued in a community-based methadone maintenance program upon release; n = 71). Entered prison treatment (p less than. 01), and completed prison treatment (p less than .001) were significantly predicted by the set of 10 explanatory variables and favored the treatment conditions receiving methadone. The present results indicate that individuals who are older in age and have longer prison sentences may have better outcomes than younger individuals with shorter sentences, meaning they are more likely to enter and complete prison-based treatment. Furthermore, implications for the treatment of prisoners with prior heroin dependence and for conducting clinical trials may indicate the importance of examining individual characteristics and the possibility of the examination of patient preference. Abstract published by arrangement with Taylor and Francis.