Power sharing is a central pillar of armed conflict mediation and de-escalation attempts worldwide and is thus employed as a tool to reduce political violence. In this study we introduce a set of novel empirical methods that aim to explicitly address the challenge of estimating the impact of power sharing agreements. We find that power sharing agreements reduce political violence in the short-term. The average effect of power sharing agreements – of the 440 agreements surveyed – is a 8% decrease in the occurrence of violence and a 18% drop in intensity of armed violence. Political power sharing provisions which are embedded in a comprehensive agreement with other power, judicial and resource-related provisions are most effective. These comprehensive agreements have an effect that is larger (a 10% and a 30% decrease in occurrence and intensity respectively) and appears to strengthen with time. We also argue that power sharing agreements can therefore provide a bridge out of the conflict trap by reducing political violence in the short-term and by strengthening the institutional protection of rights.