There is considerable variation in how armed groups fighting civil wars behave towards civilians with whom they have ongoing contact. While all armed groups use violence against civilians to some degree, some groups govern in ways strikingly similar to organized states: they extract taxes, and provide order and other public goods. Yet, others deviate from attempting to govern civilians and instead rely mostly on violence. What explains this variation? How do those armed groups that come to establish their rule decide upon such different strategies? What are the effects of armed groups' strategies on civilians' behavior? What explains civilians' decision to collaborate with a particular warring side? This project seeks to shed light onto these matters by addressing a specific phenomenon which brings to the fore the dynamics between armed groups and the civilians with whom they come into contact: the establishment of local order by armed groups within civil wars. I argue that armed groups may use violence, indoctrination or provision of stateness in order to render civilian collaboration. I explain a group's choice to use a particular strategy as the outcome of its goals, opportunities and constraints, and its expectations of civilians' responses. I approach civilian collaboration by distinguishing between spontaneous support and obedience. I identify several explanatory variables and argue that armed groups' strategies towards civilians are the most important factor in shaping collaboration. I will test my theory with qualitative and quantitative evidence from the Colombian conflict, employing a variety of sources; including a survey with Colombian ex-combatants that Stathis Kalyvas and I conducted this past year, semi-structured interviews with locals, physicians, and government officials, and a database I intend to construct using original data and existing sources. I will test observable implications of my theory by comparing across localities, armed groups, units, and individuals.