BACKGROUND Rates of obesity have risen disproportionately for ethnic minority youth in the United States. School‐based programs may be the most comprehensive and cost‐effective way to implement primary prevention in children. In this study we evaluated the effect of a school‐based obesity prevention on the outcome of body mass index percentile (BMI%), with baseline weight class and ethnicity examined as moderators. METHODS Participants (N = 125), ages 7‐11 (56% female) from 4 urban, low‐income, ethnic minority (58% black, 42% Latino) schools were recruited. Two schools received the Urban Initiatives Work to Play health intervention, and 2 demographically matched schools served as wait‐list controls. RESULTS Hierarchical multiple regression was used to analyze the independent and interactive effects of key variables on BMI%. An interaction between intervention status and ethnicity revealed Latino youth in the intervention had lower BMI% than those in the control group. Participation did not cause BMI% outcomes to decrease for black participants. CONCLUSIONS The study demonstrates the intervention is effective, but that the effectiveness varies across ethnicity. Interventions can be made more efficient and cost‐effective by targeting youth of a common ethnicity that has shown empirical responsiveness to certain program elements.