This study examines state- and regional disparities in obesity prevalence among 46,707 US children and adolescents aged 10-17 years before and after adjusting for individual socioeconomic and behavioral characteristics and area deprivation measures. The 2003 National Survey of Children's Health was used to calculate obesity prevalence in nine geographic regions and in the 50 states and the District of Columbia (DC). Logistic regression was used to estimate odds of obesity and adjusted prevalence. OLS regression was used to determine the amount of variance explained by income inequality, poverty, and violent crime rates. The prevalence of childhood obesity varied substantially across geographic areas, with the Southcentral regions of the US having the highest prevalence (> or =18%) and the Mountain region the lowest prevalence (11.4%). Children in West Virginia, Kentucky, Texas, Tennessee, and North Carolina (adjusted prevalence >18.3%) had over twice the odds of being obese than their Utah counterparts (adjusted prevalence = 10.4%). Geographic disparities in obesity were similar for male and female children. Individual characteristics such as race/ethnicity, household socioeconomic status, neighborhood social capital, television viewing, recreational computer use, and physical activity accounted for 55% of the state and 25% of the regional disparities in obesity. Area poverty rates accounted for an additional 18% of the state variance in adjusted obesity prevalence. Although individual and area level socioeconomic factors are important predictors, substantial geographic disparities in childhood and adolescent obesity remain. Prevention efforts targeting individual risk factors as well as contextual social and environmental factors may reduce geographic disparities in childhood and adolescent obesity.