Abstract Our overarching goal is to use new quantitative methods of capturing dynamic family interactions in early childhood to identify key mechanisms underlying genetic and family intervention effects on substance use and dependence (SUD) and mental health problems across adolescence. Results will demonstrate the value of observational data analyzed with modern statistical approaches and inform the specification and refinement of more potent and effective family-based interventions. Our approach builds on findings from large sample genome wide association studies to inform formation of polygenic scores that represent genetic risk for SUD and differential susceptibility to both risky and promotive family relationship dynamics. Developmental theory on gene-environment interplay has highlighted the need to move away from focusing only on simple main effect models, thus, we examine genetic association in the context of dynamic social interactions and random assignment to a family-based intervention. Although SUD is heritable, it develops and progresses within problematic family interactions and relationships. Family interventions are central to evidence-based approaches to preventing and treating SUD, and direct observation of family interactions is the most rigorous way of measuring family interactions. To date, the methodology and analysis of family interactions relevant to intervening on youth SUD relies on broad aggregate scores. The most commonly used aggregate scores that define the family interaction may simply miss pathogenic dynamics. With the advancement of statistical analyses, there is unprecedented potential for accelerating observational family research over the coming decade. The proposed study involves secondary analysis of existing videotaped observations of racially/ethnically diverse children and families from the Early Steps Multisite Trial, applying dynamic structural equation modeling and multivariate multilevel survival analysis to understand the effects of polygenic risk and family intervention on downstream adolescent SUD and mental health problems. Early Step (N=731) is a randomized trial with long-term follow-up (ages 2-19) of the effects of the family-centered intervention, Family Check-Up, on reducing problem behaviors and SUD. The extensive data include videotaped observations of children and parents across multiple contexts at ages 2, 3, 4, and 5. In addition, children were genotyped using the contemporary Affymetrix Biobank Array. Such data provide a rare and unique opportunity to utilize new statistical methods to understand early habitual family dynamics on SUD and mental health problems in adolescence. These dynamic mechanisms provide key intervention targets for enhancing intervention effectiveness and efficiency and lead to enhancing potency of family-based interventions.