Case studies have documented a solid link between protests and violence, although several mechanisms have been suggested to account for this relationship. Here, we test this connection and adjudicate between alternative mechanisms by using the tools of the large-N design. Our findings solidly support the existence of a positive relationship between mobilization and violence. Overall, we find that more popular mobilization encourages more violence, but this effect is stronger when the mobilizational wave is heading downward. This probes that low-intensity violence such as terrorism tends to spike when protests are in the waning side of the curve.