Objective To determine the efficacy of a 2-year obesity prevention program in African American girls. Design Memphis GEMS (Girls' health Enrichment Multi-site Studies) was a controlled trial in which girls were randomly assigned to an obesity prevention program or alternative intervention. Setting Local community centers and YWCAs (Young Women’s Christian Associations) in Memphis, Tennessee. Participants Girls aged 8 to 10 years (N = 303) who were identified by a parent or guardian as African American and who had a body mass index (BMI) at or higher than the 25th percentile for age or 1 parent with a BMI of 25 or higher. Interventions Group behavioral counseling to promote healthy eating and increased physical activity (obesity prevention program) or self-esteem and social efficacy (alternative intervention). Main Outcome Measure The BMI at 2 years. Results The BMI increased in all girls with no treatment effect (obesity prevention minus alternative intervention) at 2 years (mean, 0.09; 95% confidence interval [CI], –0.40 to 0.58). Two-year treatment effects in the expected direction were observed for servings per day of sweetened beverages (mean, –0.19; 95% CI, –0.39 to 0.09), water (mean, 0.21; 95% CI, 0.03 to 0.40), and vegetables (mean, 0.15; 95% CI,–0.02 to 0.30), but there were no effects on physical activity. Post hoc analyses suggested a treatment effect in younger girls (P for interaction = .08). The mean BMI difference at 2 years was –2.41 (95% CI, –4.83 to 0.02) in girls initially aged 8 years and –1.02 (95% CI, –2.31 to 0.27) in those initially aged 10 years. Conclusions The lack of significant BMI change at 2 years indicates that this intervention alone is insufficient for obesity prevention. Effectiveness may require more explicit behavior change goals and a stronger physical activity component as well as supportive changes in environmental contexts.