Economists have developed a number of theories based on warlord or bandit models to explain intra-state conflict and civil war. These models assume rational agents that agitate in a kind of institutional vacuum. This view is flawed. Ethnographic studies from civil wars suggest that livelihoods and institutions in the context of a war economy are very complex, more complex than those models suggest. We do not find an institutional vacuum – or anarchy – but a hybrid set of overlapping and contradictory sets of rules. This paper applies several concepts of institutions discussed in literature on new institutional economics and sociology to an analysis of the emergence and logic of the rules of the game in the political economy of civil wars. The analysis indicates that contracting in civil wars, whether complete or incomplete – and the opportunity to grab (Skaperdas), to loot (Collier) and to exploit others (Hirshleifer) – takes place on many different scales and between different agents, not only among combatants. This creates a complex, dynamic and hybrid institutional amalgam of coercively imposed rules, traditional norms and co-existing formal institutions.