Competing hypotheses are tested on an extensive set of defense-burden data to determine the general factors that influence states' levels of military spending. Results provide some clear answers to longstanding questions and supply new findings that beg further investigation. When controls are introduced for domestic political and economic factors, several international factors, including alliances and rivalries, lose statistical significance. Consistent with liberal theory, regime type has a robust effect: democracies spend proportionately less on defense than other states. As implied by realism, under conditions of economic growth or high levels of wealth, "extra" resources are diverted disproportionately to the military.