This is an analysis of the spatial structure of political participation in the United States using spatial econometric techniques and newly available geo-coded data. The results provide strong evidence that political participation is geographically clustered, and that this clustering cannot be explained entirely by social network involvement, individual-level characteristics, such as race, income, education, cognitive forms of political engagement, or by aggregate-level factors such as racial diversity, income inequality, mobilization or mean education level. The analysis suggests that the spatial structure of participation is consistent with a diffusion process that occurs independently from citizens' involvement in social networks.