Power-sharing arrangements aim to reduce the risk of civil conflict by guaranteeing potentially warring parties a role in a country’s government, thus lessening the stakes of political contestation. In this way, power-sharing reduces the risk that spoilers will resort to violence if they do not succeed in the process of democratic electoral contestation. While power-sharing can reduce the incentive of electoral losers to renege on their commitment to democracy, we argue that this depends on the nature of the relevant groups, as well as on the political institutions that are chosen. The degree to which power-sharing agreements are able to promote civil peace thus depends in part on the relative military capacity of the fighting parties, as well as on the potential role of ‘spoilers’. The ideal environment for power-sharing to shape peace is when the sides are evenly balanced and the costs of war are relatively high. In contrast, when groups are less evenly matched and the costs of war low, power-sharing implies nonproportional distributions of power and positive incentives for spoilers. Under such conditions, power-sharing may increase rather than reduce the risk of civil conflict.