The proposed study tests the contribution of three key environmentally- mediated variables to male delinquency. These are: the extent to which the boy is well supervised/monitored, how well his caretakers use consistent, predictable, well-suited discipline methods, and the amount of positive reinforcement, encouragement, and help he receives. Subjects are 120 male juvenile offenders who are randomly assigned to one of two conditions: Specialized Foster Care (i.e., the experimental group) or Group Care Centers or Homes (i.e., the control condition). It is expected that in both conditions varying levels of the three mediating/intermediate variables of interest, as well as another well- established predictor of delinquency, association with antisocial peers, will be present. Multimethod/multiagent measures will be used to form indicator and construct scores of the hypothesized mediators (i.e., Monitoring, Discipline, and Positive Reinforcement) and outcomes (i.e., Association with Deviant Peers and Delinquency). Subjects will be assessed at baseline, 2 and 4 months after placement, at termination, and at 6-month follow-up. Major hypotheses are (1) that subjects in the experimental condition will receive more intensive monitoring, consistent discipline, and positive reinforcement, and (2) that regardless of group assignment, those subjects with higher scores on the three hypothesized mediators will have less association with deviant peers and less delinquency.