Political scientists repeatedly find that a person's turnout in past elections strongly predicts her future turnout. Scholars often interpret the association between past and future turnout as evidence of "habitual voting." However, the discipline has no widely accepted conceptual definition or empirical measure of habitual voting, and few systematic investigations of the causes and consequences of habitual voting. In this dissertation, I draw on research in psychology to propose a new theoretical framework for understanding turnout focusing on habit. The framework describes the processes that govern the formation of voting habits over time, the defining features of voting habits, and the factors that "make" and "break" voting habits. I test the propositions in the dissertation using data from an original panel survey of Minnesota registered voters conducted before and after the 2010 election in conjunction with state information on respondents' turnout over a six-year period. The data are first used to develop and validate a direct measure of the psychological features of voting habits. I use the measure to investigate the causal relationships among voting over time, changes in mood, efficacy, and geographic residence, and the strength of voting habits, among other factors. In the final chapters of the dissertation, I examine the different effects of mobilization messages and early voting policies on habitual and non-habitual voters.