This study analyzes the social and economic correlates of air pollution exposure in U.S. cities using a unique dataset created as a by-product of the EPA’s Risk-Screening Environmental Indicators model and finds evidence of disproportionate exposure to environmental hazards in communities with higher concentrations of lower-income people and people of color. We improve on previous studies of environmental inequality in three ways. First, where previous studies focus on the proximity to point sources and the total mass of pollutants released, our measure of toxic exposure reflects atmospheric dispersion and chemical toxicity. Second, we analyze the data at a fine level of geographic resolution. Third, we control for substantial regional variations in pollution, allowing us to identify exposure differences both within cities and between cities. We combine 1998 data on toxicity-adjusted exposure to air pollution with 1990 Census block group data for urbanized areas. We find that blacks tend to live both in more polluted cities in the U.S. and in more polluted neighborhoods within cities. Hispanics live in less polluted cities on average, but they live in more polluted areas within cities. We find an extremely consistent income-pollution gradient, with lower income people significantly more exposed. Our findings highlight the importance of controlling for interregional variation in pollution levels in studies of the demographic correlates of pollution.