Recurrent episodes of civil unrest significantly reduce the potential for economic growth and poverty reduction. Yet the economics literature offers little understanding of what triggers civil unrest in society and how to prevent it. This paper provides a theoretical analysis in a dynamic setting of the merits of redistributive transfers in preventing the onset of (and reducing) civil unrest and compare it with policies of more direct intervention such as the use of police. We present empirical evidence for a panel of Indian states, where conflict, transfers and policing are treated as endogenous variables. Our empirical results show, in the medium-term, redistributive transfers are both a more successful and cost-effective means to reduce civil unrest. Policing is at best a short-term strategy. In the longer term, it may trigger further social discontent.