This project longitudinally tests the efficacy of school-based indicated and universal preventive intervention strategies, beginning at school entry, to reduce risk for early-onset conduct disorder, substance abuse, and school failure. The design includes an indicated sample at high risk (children showing early cross-setting aggressive behavior at school entry) and a selected sample at lower risk (children residing in socioeconomically disadvantaged neighborhoods). A social interactional perspective emphasizing the role of coping-competence in development drives the interventions and design. Hypothesized risk factors are: negative classroom experiences, disturbed peer relations, parenting difficulties, history of child temperrament difficulties, family adversity. The putative protective factors are positive school climate and family social support. Critical information is sought about the ontogeny and prevention of conduct disorder and related outcomes, particularly through the inclusion of the previously understudied population of high-risk and lower-risk females who compose 40% of the sample. For the high-risk sample, two intervention conditions are compared experimentally: (1) a multicontextual preventive intervention (classroom, peer, reading-tutoring, family), and (2) universal school- wide conflict management program. The piggyback design for the lower- risk sample compares the school-wide conflict-management program to the first-grade classroom program. Salient design features include: (a) matched random assignment of schools; (b) sufficient power to use school as the unit of analysis if necessary; (c) multistage selection of high- risk youth; (d) multi-method, multi-source longitudinal measures of risk and protective factors, proximal outcomes related to coping-competence and aggression, and distal outcomes; (e) systematic observational assessment of parent-child interaction; and (f) promotion and verification of fidelity. The project capitalizes on data-reduction and covariance techniques, multivariate growth curve analysis, and logistic regression in evaluating outcomes (e.g., conduct disorder, antisocial acts, early substance use), addresses the systematic prevention of conduct disorder, tests the theoretical foundation of the school-based interventions, and blends developmental and epidemiological perspectives.