OBJECTIVES: Cox regression is used to analyze relapse patterns of adolescents treated for psychoactive substance use disorder (PSUD). The objective is to evaluate the role numerous psychosocial, treatment and environmental characteristics play in the relapse process in this treatment population. It is clear that the PSUD disease and recovery process are unique in adolescents and that relapse and recovery need rigorous study. Relapse is the most important treatment outcome. METHODS: Subjects are 509 adolescents discharged from an ASAM-defined Level 1.A primary inpatient treatment program from 2001-2005. Data was collected as part of the treatment program's annual outcomes evaluation. The sampling frame was all who successfully completed treatment. Response rate was 61%. Analysis of characteristics of nonrespondents showed no significant differences compared to respondents. The survey is based on a 230-item questionnaire. Treatment records of each adolescent completing the questionnaire were obtained for matching treatment outcomes from the questionnaire to treatment and sociodemographic variables contained in treatment records. A comprehensive data set was created from these two sources. Data were analyzed using Cox proportional hazard regression. RESULTS: Results indicate race (Whites were 59.2% less likely to relapse than other races; blacks are 4.9 times more likely.), gender (males 1.28 times more likely), participation in support groups (participants 23.7% less likely), school attendance (attendees 21.6% less likely), supportive friendships (one SD change on scale corresponds to a 7.7% reduction in relapse risk), and cannabis dependence (cannabis diagnosis 28.7% more likely) are significant determinants of relapse, certis paribus. CONCLUSIONS: Several risk factors for relapse are identified that can be addressed in primary treatment. For instance, treatment programs emphasizing friendships skills by application of social cognitive theory might be considered.