Why do some protests turn violent while others do not? The violent escalation of demonstrations is subject to massive media coverage, yet little empircal research on the causes of this has been done. This thesis offers a spatially and temporally disaggregated research design that analyzes the dynamics of protest events in Africa and Asia from 2010-2018. The results from various specifications of the OLS regression model show that government repression of nonviolent protest preludes violent escalation. Repression of violent protest has the opposite effect, instigating violent protesters to switch to nonviolent means. Furthermore, I find that the association between repression and violent escalation is conditional upon whether nonviolent protests are organized or not. Even in the face of repression, organized nonviolent protests are more effective at maintaining nonviolent discipline than their disorganized counterparts.