This research focuses on the forms of exclusion that democratizing processes have historically facilitated. The dynamics of democratization often lead political coalitions to change electoral rules to simultaneously extend and constrict the right to vote across different categories of persons, as well as to reinforce existing exclusions. This pattern occurred in all the 'exemplary models' of early democratization, and yet the historical narratives relied on by the comparative democratization literature neglect its exclusionary dimension, and thereby misinform comparative theory building. The dissertation empirically documents the "dark side of democratization" in the three paradigmatic cases of the United States, the United Kingdom, and France, and develops and tests a theory explaining cross-national and cross-time variation. At key moments in a country's development, political entrepreneurs advance ideas of community belonging for the purpose of securing a governing coalition. When successful the ideas of political community are embedded in new institutions and in public opinion, shaping the expectations of political agents across the political spectrum and resulting in higher costs of coalition-building and political mobilization across categories of people. The exclusions were thereby made resilient to subsequent democratizing processes. The dissertation advances research the role of ideas in social science by focusing on the micro-foundations of democratic exclusion. The model predicts various of political behavior that are integrally important to democratization, and is tested against debates, voting behavior, and correspondence in and outside of parliaments, legislatures, and constitutional conventions. The data draws on archival field work research, multiple datasets of legislator behavior, constituency demographics, and institutional change. These allow for the identification of stable patterns as well as change across time, and supplement a process tracing research design.