Compared three different methods for teaching social decision making and problem solving skills to elementary school children: a standard social problem solving curriculum, a discovery learning condition, and a directed learning approach. Gains on specific indices of problem solving were noted for all treatment conditions, but generalization to affective and behavioral indices 3 months after the end of the intervention was greater for the discovery and directed conditions. The interaction of instructional method by initial adjustment level of children-at-risk, average, and competent-showed broad-based benefits of the social problem solving programs, especially to at-risk students. The discussion highlights the implication of the findings for the delivery of school-based preventive social competence/social decision making programs.Psychologists planning preventive interventions with children show an increasing preference for social-cognitive problem solving methods (Denham & Almeida, 1987;Durlak, 1983). Social decision making and problem solving have been documented to be central to the interpersonal and emotional adjustment of children and adolescents (Benard, Fafoglia, & Perone, 1987;Elias, 1990b;Michelson, Sugai, Wood, & Kazdin, 1983). Many behavioral and emotional difficulties can be conceptualized as deficiencies in component skills of social decision making and problem solving. These include problems with defining appropriate goals, lack of alternative thinking, poor formulation and implementation of plans, and failures to modify plans in light of experiential feedback (Kendall & Hollon, 1981; Rutter, 1987).We would like to acknowledge the support and assistance of many colleagues, particularly