Recent work in comparative politics and international relations has shown a marked shift toward leaders as the theoretical unit of analysis. In most of the new theoretical models a core assumption is that leadersW hen testing theories, researchers would like the theoretical and empirical units of analysis to match. In international relations, however, this is often not the case. On the one hand, neo-realists and others who focus on the system as their unit of analysis as well as scholars who base their theories on the unitary rational actor assumption usually rely on the statements and memoirs of leaders in their empirical research (Copeland 2000). Whereas their theoretical unit of analysis is the system or the state as a unitary rational actor, their empirical unit of analysis is the leader. On the other hand, scholars of diversionary war and, recently, the democratic peace employ leaders as their theoretical unit of analysis. In their statistical research, however, these scholars tend to rely on the country or country-year as the empirical unit of analysis (Gelpi 1997;Leeds and Davis 1997;Reiter and Stam 2002). Such mismatches between the theoretical and empirical unit of analysis can significantly weaken empirical tests and produce faulty inferences. A fundamental hurdle for scholars who focus on leaders as their theoretical unit of analysis has been the lack of appropriate data on leaders. Earlier drafts of this article were presented at the PIPES seminar at the University of Chicago, the University of Virginia, and the Annual Convention of the American Political Science Association, Washington D.C. We thank the commentators at those presentations; in particular we thank Thomas Blackwell, Chris Gelpi, Jessica Hardesty, Bob Keohane, Craig Koerner, Jeff Legro, Thomas Lumley, Kevin Morrison, Davide Raggi, Dan Reiter, Sebastian Rosato, Angelo Secchi, Randolph Siverson, Alastair Smith, David Soskice, Allan Stam, and Terry Therneau. Mistakes, omissions, and other assorted infelicities are our own responsibility. Authors' names are in alphabetical order. theories on the appropriate empirical unit of analysis and to promote further theoretical work on the role of leaders in international and comparative politics, we introduce a new data set of all leaders between 1919 and 1999.Recent work in international relations and comparative politics attempts to build powerful new theories on a solid methodologically individualist basis by focusing on the incentives and constraints of leaders (Bueno de Mesquita et al. 2003;Cheibub and Przeworksi 1999). A central assumption in much of this new research is that leaders act to stay in power (Downs 1957). Very little is empirically known, however, about the factors that affect the tenure of leaders (Bienen and van de Walle 1991;Bueno de Mesquita et al. 2001). To provide some baseline, this article analyzes the new data to assess how a broad range of domestic and international factors affect the tenure of leaders.To promote further theoretical and empirical research on the role of ...