Objective. Debate about market–oriented school–choice proposals often centers on questions of whether they will help or hurt minorities and the poor. We examine the locational decisions of different types of charter schools in the District of Columbia (D.C.) to assess their distributional consequences. Methods. We employ ordered probit regression to estimate models of the degree to which census tracts are served by charters. Results. Charters are more likely to locate in areas with high proportions of African–American and Hispanic residents than in the predominantly white neighborhoods, and more likely to locate in neighborhoods with middle incomes and high home ownership than in either poor or wealthy areas of the city. This is especially true of those operated by for–profits and those chartered by the elected rather than appointed chartering body. Additionally, we observe charters taking political and practical considerations into account when deciding where to locate. Conclusions. Proponents claim that charter schools will locate where need is greatest, while critics fear they will shy away from neighborhoods housing disadvantaged and minority students. We find that both camps are oversimplifying. Locational patterns are more complex and appear to be sensitive to variations in the type of charter school as well as the institutional characteristics of the chartering agency. Although market incentives are important, so too are pragmatic factors and institutional context.