Over twenty years of systematic empirical research has been unable to substantiate a relationship between regime type and war–proneness. Using an improved statistical model, I demonstrate that from 1960 to 1980, democratic nations were less involved in military conflict than other regime types. Estimates of this relationship are robust to different operational definitions of both war and democracy and to adding control variables for other possible correlates of war. Further tests of categorical associatio n confirm these results, in addition to suggesting why previous efforts have failed to establish the relative pacifism of democracies.