Background: The adverse consequences of adolescent alcohol use are substantial and varied. Community-based approaches to prevention have gained favor over the past 30 years, and the use of coalitions has become a popular model by which to plan and implement interventions. Although theory suggests that coalition capacity and community readiness are likely to affect the quality of implementation and the efficacy of the interventions, empirical work in this regard is lacking. Methods: Using data from an evaluation of the Vermont Strategic Framework State Incentive Grant, a theory-based multiple mediation model was tested that examined the direct effects of coalition capacity and community readiness, and the mediated effects of intervention comprehensiveness, evidence base, and fidelity of implementation, on past-month alcohol use and binge alcohol use among high school students in 24 intervention communities. Coalition and community member surveys were used to collect data on coalition characteristics and community readiness, data from progress reports were abstracted to measure implementation characteristics, and Youth Risk Behavioral Surveillance System survey data was used to measure past-month use. It was hypothesized that greater levels of coalition capacity and community readiness would lead to greater reductions in alcohol and binge alcohol use over time and that implementation characteristics would mediate these relationships. Results: Significant effects of coalition capacity on alcohol use were found, although no significant mediators of this relationship emerged. There were no significant effects on binge alcohol use. Exploratory analyses indicated that the total number of interventions implemented significantly mediated the relationship between readiness and reductions in binge alcohol use. Discussion: This study provides an empirical test of theoretical relationships commonly proposed in the community-based substance use prevention literature. One hypothesis was supported and the results of exploratory analyses identified a new, potentially important mediating factor. Improvements in measures and the application of the proposed mediation models to larger studies are needed to improve our understanding of the mechanisms operating in communities that produce behavioral change.