In the past decade, violent protests have been on the rise. We have seen riots not only against authoritarian regimes, as in the Arab Spring, but also defying reforms or structural problems in consolidated democracies, as we witnessed with the Yellow Vest protests in France or the George Floyd protests in the United States of America. However, violent protests are more common in weak democracies, which combine civil and political rights with ineffective institutions. Weak democracies have several incentives for violent protests and yet, these events vary not only across but within countries. How do we explain that some regions are more or less likely to shape violent protests within a weak democracy?Scholars have mostly focused on whether the protesters or the state explain violent protests. And the ones that pay attention to the interaction, generally look only at the effects of state repression. From a relational approach, my dissertation proposes a theory of violent protests that focuses on state accommodation capacity and on how the interaction of this capacity, through conflict prevention agencies, and civil society organizations generate higher or fewer chances to use violent repertoires. My argument is that subnational regions with robust conflict prevention agencies and structured civil society organizations will be less likely to have violent protests. The interaction of these two factors generates the mechanisms needed to institutionalize a dialogue dynamic that may not be exempt from protests, but it is from the use of violence. We need to pay attention to state accommodation capacity.I develop my argument by comparing the likeliness of violent protests in four regions of post-Fujimori Peru, from 2001 to 2018. With ineffective institutions, which implies a weak state and a poor representation, Peru should have a higher rate of violent protests. However, there are regions where the use of violent repertoires is less frequent. Employing interviews, an original database of protests from 2008 to 2018, and archival data from local newspapers and reports in Arequipa, Puno, Lambayeque, and Cajamarca, I show that when robust local CPAs and a non-competitive ecology of CSOs are in place, the likeliness of violent protest will reduce. Conversely, when one of these two factors is missing, we will see that the mechanisms that deactivate violence will not function and multiple actors will have more incentives to radicalize/repress.