Adolescent drug use remains a major health and safety problem. The dilemma for administrators and policy planners is to determine what types of prevention programs to implement, given the contingent nature of evidence regarding program effectiveness. Although the community-based prevention has emerged as a viable way of reducing the risk of drug use among adolescents, it has been unclear if it is more effective than other approaches, such as school-based and family programs. This paper uses systematic review methodology to analyze the findings of 51 studies investigating the effectiveness of 65 community-based prevention programs representing more than 11,000 subjects at baseline. Recent moves in academic and policy circles to strengthen the social science research evidence base have suggested a need for systematic research synthesis and for greater accessibility of sound research evidence. As an alternative to designing an evaluation, meta-analysis can extract from the body of existing literature a reasonably credible assessment of the impact of interventions that have been tried. Using a standardized procedure, it permits to calculate the effect sizes for drug use measures reported in each study. The effect size provides, in standard deviation units, a uniform measure of quantitative differences in drug use outcome. The measurements are comparable across studies and hence subject to further analysis such as assessing the overall magnitude of effect produced by a given type of intervention. This paper focuses on contextual factors in explaining alcohol, tobacco, and drug use outcomes. Program context includes aspects of the target populations and the settings in which the study is conducted. The optimal setting for delivering prevention services to youth has long been a matter of debate among researchers and practitioners. In the present work, community-based interventions are classified by the main channels used for program delivery and by the program focus ("community-only", "school-community", and comprehensive programs). The paper also assesses the relative effectiveness of "school-only" programs by building upon earlier meta-analyses of the school-based drug education program evaluations (Tobler et al., 2000; Roona, Streke, & Marshall, 2003). Using similar meta-analytical procedures, the present study employs a more disaggregated approach together with some innovative statistical techniques that more reasonably deal with data problems and distributions underlying drug use. Overall, the findings provide evidence of a modest benefit of program participation for the average youth. The results lend some support to the proposition that comprehensive interventions operating through multiple channels are more effective than individually focused, single-channel interventions in influencing drug behavior throughout youth populations. The programs with a broader focus delivered in multiple settings ("comprehensive" and "school-community") tend to outperform more narrowly focused programs delivered in a single setting ("community-only" and "school-only") on major substance use outcomes. School-based programs and community interventions appear to have a stronger impact when they work in tandem rather than as separate, stand alone interventions. The differential effectiveness of prevention programs may be attributable to the particular opportunities and constraints that characterize each program setting, because the setting can shape the type, style, and amount of service offered.