This is a longitudinal study of three birth cohorts of youngsters who were considered at risk because of anti-social behavior or because of officially recorded delinquency at early ages. The study followed a sample of 245 boys in the fourth, seventh, and tenth grades in 1980 (Part 1) and again in 1985 (Part 2). Two screening devices, or "gatings," were used to predict future delinquency. The first procedure, triple gating, was based on teachers' ratings of school competence, mothers' reports of anti-social behavior in the home, and parental monitoring. The second procedure, double gating, used only the teachers' ratings and mothers' reports. Data were collected on the boys' family, school, and criminal backgrounds. Variables include measures of independence and achievement, family criminality, home conduct problems, school disruptiveness, school competence, parental authoritarianism, parental conflict, self-reported delinquency, peer delinquency, and drug and alcohol use.