Longitudinal studies are needed to increase understanding of the causes of childhood obesity. To identify 1- and 2-year predictors of excess weight gain among preadolescents, the authors conducted a prospective cohort study of fourth- and fifth-grade students in 16 elementary schools located in multiethnic, low-income neighborhoods in Montreal, Quebec, Canada, that were participating in the evaluation of a school-based heart health promotion program. Subjects included 2,318 children aged 9–12 years with baseline and 1-year follow-up data and 633 children aged 9–11 years with baseline and 2-year follow-up data. One-year predictors of highest decile of change in body mass index (BMI) identified in logistic regression analyses included baseline BMI of 90th percentile or more (odds ratio (OR) = 2.66, 95% confidence interval: 1.80, 3.94) in boys and baseline BMI of 90th percentile or more (OR = 2.34, 95% confidence interval: 1.46, 3.76), no sports outside school (OR = 1.90, 95% confidence interval: 1.18, 3.06), and playing video games everyday (OR = 2.48, 95% confidence interval: 1.04, 5.92) in girls. Two-year predictors included baseline BMI of 90th percentile or more (OR = 3.26, 95% confidence interval: 1.52, 7.01), no sports outside school (OR = 2.14, 95% confidence interval: 0.96, 4.77), and least active (OR = 2.18, 95% confidence interval: 1.01, 4.71) in boys; only baseline BMI of 90th percentile or more (OR = 2.22, 95% confidence interval: 1.02, 4.81) was significant in girls. Results suggest the need for interventions to promote increased physical activity in children. Am J Epidemiol 2000;152:739–46.