The authors address the question of state militarization under conditions of ethnic and other diversity. "Primordialist" claims about ancient hatreds, fear, and insecurity in such societies would lead one to expect that fractionalization, polarization, and ethno-nationalist exclusion would prompt governments to militarize heavily. But contrary to such expectations, the authors find that higher levels of ethnic diversity predict lower levels of militarization, whereas higher polarization and ethno-nationalist exclusion trigger neither lower nor higher levels of militarization. If fractionalization lowers the hazard of civil war, as many find, then it does not happen by way of a "garrison state" effect. The authors discuss two potential explanations for their findings, one drawing from the empirical conflict literature, the other stemming from economists' study of public goods provision under conditions of diversity. They argue that their findings are best seen as consistent with and complementary to the empirical literature on conflict onset and duration.