This paper develops a unified approach to study participation and voting in multiple elections. The theoretical setting combines an “uncertain-voter” model of turnout with a spatial model of voting behavior. We apply our framework to the study of turnout and voting in U.S. presidential and congressional elections. We structurally estimate the model using individual-level data for the 2000 elections, and quantify the relationships between observed individual characteristics and unobserved citizens’ ideological preferences, information, and civic duty. We then use the estimated model, which replicates the patterns of abstention, selective abstention, split-ticket voting, and straight-ticket voting observed in the data, to assess the effects of policies that may increase citizens’ information and sense of civic duty on their turnout and voting behavior.