This study summarizes, using meta-analytic techniques, results from 94 studies of school-based prevention activities that examined alcohol or other drug use outcomes. It set out to determine what features of school-based substance abuse prevention programs are related to variability in the size of program effects, It asked (1) Which populations (e.g., high risk vs. general population) should be targeted for prevention services? (2) What is the best age or developmental stage for prevention programming? (3) Does program duration matter? and (4) Does the role of the person delivering the service (e.g., teacher, law enforcement officer, peer) matter? The results suggest that targeting middle school aged children and designing programs that can be delivered primarily by peer leaders will increase the effectiveness of school-based substance use prevention programs. The results also imply that such programs need not be lengthy. The evidence related to the targeting issue is sparse, but suggests that, at least for programs teaching social competency skills, targeting higher risk youths may yield stronger effects than targeting the general population. Suggestions for future research are offered.