Extant theories of the effect of domestic politics on international conflict hold that democracies are transparent and very selective in choosing when to go to war. If these theories are correct, it is difficult to explain why wars involving democracies occur at all as well as why democracies sometimes find themselves fighting prolonged, politically divisive wars. To explain these puzzles, I argue that we need to move past static conceptions of accountability (which is assumed to be constant within and across democracies) and resolve (which is often seen simply as the willingness to fight or not). I develop two game-theoretic models illustrating how domestic political processes within the state (such as party competition, the popularity of the leader, and the electoral cycle) can produce variation in the degree to which leaders expect to be held accountable for foreign policy outcomes, and the implications such variation has for the level of resolve the leader will exhibit in international crises. I derive several novel predictions regarding the duration of interstate conflict, the timing of low-level disputes, the outcomes of conflicts, and the domestic political consequences of international outcomes. Statistical analyses provide at least qualified support for nearly all of the theoretical expectations. Several illustrative cases for each theoretical model are discussed. I conclude that scholars of international relations would do well to view accountability as the outcome of dynamic political processes taking place within the state, rather than characteristics of institutional arrangements.