If nonviolence is the key to success, why do excluded groups resort to violent protests more than nonviolent protests? Why are nonviolent resistance campaigns by ethnic groups frequently met with state repression? In this paper, I argue that nonviolent campaigns do not work for excluded groups because they are often met with extreme state repression, which leaves no option for excluded groups but to resort to violent campaigns. Drawing on data from the Nonviolent and Violent Campaigns and Outcomes (NAVCO), and the Ethnic Power Relations (EPR) datasets, I find that state repression tends to be the most intense in nonviolent campaigns led by marginalized groups. I further explore repression dynamics by analyzing data from India, where I find that repression of nonviolent protests tends to be the most fatal in the state of Jammu and Kashmir – the only state where Muslims are a majority. Analyzing Facebook data from the same time period illustrates how negative stereotypes associating Kashmiri people with violence and terrorism can legitimate state repression of minority groups. These empirical findings challenge the conventional wisdom that nonviolent campaigns are likely to be more successful than violent campaigns. The critical aspect missing in the civil resistance literature is the link between state repression, nonviolent resistance, and the ethnic identity of the protesters.