Background: This study examines whether the extent of states' physical education (PE) requirement policies and the prevalence of schools requiring PE are associated with children's physical activity (PA) and obesity. Methods: Two cycles (2002-2003 and 2006-2007) of cross-sectional data with individual- and state-level information were analyzed. Information on states' PE requirement was from the 2003 and 2006 Physical Education-Related State Policy Classification System, and the prevalence of schools requiring PE was from the 2002 and 2006 School Health Profiles. The individual-level sample includes 11- to 17-year-old children from the 2003 (N = 25,251) and 2007 (N = 23,728) National Survey of Children's Health. The associations of states' PE requirement and schools' PE requirement with children's PA and obesity were examined, controlling for individual, family, and state covariates. Variation in the associations by individual and family characteristics was also tested. Results: States' PE requirement was not associated with any outcome. In cycle 1, overall, a 10% increase in the percentage of schools requiring PE was associated with a 28% increase in the number of days having vigorous PA per week. In cycle 2, the association was not significant. However, significant variation in the association by gender in cycle 2 suggests an influence of schools' PE requirement on girls' PA only. No association was found between schools' PE requirement and obesity. Conclusions: Gaps exist between state PE-related policies and implementation in schools. However, schools' PE requirement seems to improve children's PA with some gender variation. The association between schools' PE requirement and children's weight is less clear. (Contains 5 tables.)