In Thailand, India, Libya, and elsewhere, governments arm the populace or call up volunteers in irregular armed groups despite the risks this entails. The widespread presence of these militias, outside the context of state failure, challenges the expectation that governments uniformly consolidate the tools of violence. Drawing on the logic of delegation, we resolve this puzzle by arguing that governments have multiple incentives to form armed groups with a recognized link to the state but outside of the regular security forces. Such groups offset coup risks as substitutes for unreliable regular forces. Similar to other public-private collaborations, they also complement the work of regular forces in providing efficiency and information gains. Finally, these groups distance the government from the controversial use of force. These traits suggest that militias are not simply a sign of failed states or a precursor to a national military, but an important component of security portfolios in many contexts. Using cross-national data (1981–2005), we find support for this mix of incentives. From the perspective of delegation, used to analyze organizational design, global accountability, and policy choices, the domestic and international incentives for governments to choose militias raise explicit governance and accountability issues for the international community.