The extent of local order varies widely in rebel held areas, from total chaos to well-run governing institutions. When these institutions exist, why do some include and even empower civilians to run community affairs, while others exclude civilians from governance? I argue that rebels choose different governing strategies that maximize their utility of territorial control, based on certain characteristics of civilian inhabitants populating the territory. Rebels’ constituency determines whether rebels seek to govern civilians or control them solely with coercive violence, and community cohesion (or lack thereof) then determines the type of institutions that rebels develop. I focus on three different outcomes for communities under rebel control—no institutions, exclusive institutions, and inclusive institutions. I test my argument using historical, statistical, and case evidence, leveraging original cross-national data on local order in rebel held territory as well as interviews with village heads, ex-combatants, and community members in Aceh, Indonesia. The results provide support for my theory and yield implications for our understanding of human security during conflict and the determinants of civilians’ political and social reality during war.