Male and female managed care clients randomized to day hospital (n=154) or community residential treatment (n=139) were compared on substance use outcomes at 6 and 12 months. To address possible bias in naturalistic studies, outcomes were also examined for clients who self-selected day hospital (n=321) and for clients excluded from randomization and directed to residential treatment because of high environmental risk (n=82). American Society of Addiction Medicine criteria defined study and randomization eligibility. More than 50% of followed clients reported past-30-day abstinence at follow-ups (unadjusted rates, not significant between groups). Despite differing baseline severities, randomized, self-selecting, and directed clients displayed similar abstinence outcomes in multivariate longitudinal models. Index treatment days and 12-step attendance were associated with abstinence (p<.001). Other prognostic effects (including gender and ethnicity) were not significant predictors of differences in outcomes for clients in the treatment modalities. Although 12-step attendance continued to be important for the full 12 months, treatment beyond the index stay was not, suggesting an advantage for engaging clients in treatment initially and promoting 12-step attendance for at least a year.