The phenomenon of suicide attacks has dramatically expanded over the last twenty years, rising from no events in 1980 to a total of 1,398 events by 2008. A prominent theory has argued that suicide attacks are a coercive strategy aimed at ending foreign military occupation by democracies. Yet these conclusions are based on a research design that is affected by selection bias and that fails to distinguish foreign occupations from cases of groups seeking independence or autonomy, which we term domestic occupations. Analyzing an original data set that distinguishes the different types of occupation, we find that only foreign occupations have a strong and consistent effect on the incidence of suicide attacks. The reason, we argue, is that suicide attacks only become cost effective when targets are both hardened and accessible, a strategic environment that is more common to civil wars and foreign occupations than to domestic occupations.