Does the local, racial context influence racial differences in culture? I answer this question by testing predictions from group threat theory and the cultural division of labor about which high schools have greater black-white differences in basketball performance. Data are from the National Education Longitudinal Study are analyzed with multilevel ordered probit models. After controlling for predictors of sports performance in students families, schools, and neighborhoods, we find evidence for both theories. Black-white differences in basketball performance is greater in schools that are about 50% black, as group threat predicts, and in schools with more hierarchical segregation within them, as the cultural division of labor predicts. We also find that racial conflict within the schools mediates the effect of group threat. The theoretical implications of the findings are discussed.