Background: The Theory of Triadic Influence (TTI) provides a comprehensive framework for understanding adolescent substance use. Objectives: We examined mechanisms by which a TTI-guided social-emotional and character development program, "Positive Action" (PA), influences adolescent substance use. Study data come from the PA-Chicago, longitudinal matched-pairs cluster-randomized control trial. A diverse, dynamic cohort of approximately 1,200 students from 14 low-performing schools were assessed at eight points of time, between grades 3-8, across a six-year period. Students completed scales related to substance use, self-control, deviant peer affiliation, and school attachment, adapted from the Risk Behavior Survey, Social-Emotional and Character Development Scale, Conventional Friends Scale, and People in My Life Scale. After testing the overall effect of PA on substance use, we used latent growth modeling to assess whether effects on each outcome were mediated by longitudinal changes in three composite measures aligning with the TTIs three streams. Results: Students in PA schools reported fewer experiences with drinking, getting drunk, and overall substance use. In the multiple mediator models, significant indirect effects of PA on substance use via changes in self-control were evident. Conclusions/ Importance: Findings are consistent with theory and past research suggesting the influence of self-control on youth substance use. Future studies should include implementation in different settings and additional theory-based measures.