The state repression literature has paid disproportionate attention to explaining and predicting repression in response to violent dissent while overlooking the role of nonviolent dissent in the repression-dissent process. I present a theory of repression that considers how the tactics and participants involved in nonviolent dissent uniquely threaten political power and affect leaders' decision-making differently than violent dissent. Violent dissidents rely on physical coercion and must possess the willingness and ability to engage in violence; this requirement tends to result in homogeneous dissident groups that are not representative of the larger population. Nonviolent dissidents, on the other hand, use non-physically coercive tactics and are often more representative of the general public than violent dissidents. These peaceful tactics and diverse participants increase the chance that the public will disagree with the use of repression and sanction leaders. Strategic leaders recognize that repression of nonviolent dissent may backfire in this way, but they cannot easily gauge the likelihood that audiences will learn about or be angered by the use of repression against nonviolent dissidents. I suggest that certain domestic institutions provide leaders with information on the likelihood of backfire that serves as a constraint on repressive behavior. Specifically, I predict that leaders in those states with high levels of press freedom and a constitutionally protected right of petition expect that repression of nonviolent dissent is more likely to backfire and will be less likely to repress as a result. A free press threatens to disseminate information on the state's use of force against peaceful citizens, and the potential media attention increases the likelihood that citizens learn about the abuse and sanction leaders. Where the right of petition is constitutionally protected, leaders have an additional expectation that repressing nonviolent dissent will upset domestic audiences and that these angry citizens will mobilize to sanction repressive leaders. I test these predictions using the Nonviolent and Violent Campaigns and Outcomes (NAVCO) v2.0 dataset and use an illustrative case of repression in Mexico to test the plausibility of the theory.