Meta-analyses of skills-based prevention interventions show moderate effect sizes for increasing social competencies and decreasing behavior problems. While the literature suggests that prevention programs can be effective, rigorous independent research is lacking regarding the efficacy of many specific programs. The present study is based on a randomized-control experiment evaluating Second Step in 12 Maryland public elementary schools to assess the effects of the program on parent and teacher ratings of student behavior. Ratings using the Social Competency Rating Form had previously been considered as a single global measure of student behavior, and had not been found to be affected by the intervention. Nonetheless, a re-consideration of the psychometric properties of the scale and its sensitivity to skills taught by the Second Step curriculum led to the speculation that separation of the global measure to reflect distinct ratings of social competency and problem behavior might reveal effects on the social competency component. Analyses show no effects on parent or teacher ratings of social competency or on teacher ratings of problem behavior. In some analyses, students in treatment schools had nearly twice the odds of being classified in a "problem" group according to ratings made by their parents than did students in control schools. Results were supported by sensitivity analyses using weights and imputation.