Abstract: From the streets of Hong Kong to Ferguson, Missouri, civil disobedience has again become newsworthy. What explains the prevalence and extremity of acts of civil disobedience? This paper presents a model in which protest planners choose the nature of the disturbance hoping to influence voters (or other decision-makers in less democratic regimes) both through the size of the unrest and by generating a response. The model suggests that protesters will either choose a mild “epsilon” protest, such as a peaceful march, which serves mainly to signal the size of the disgruntled population, or a “sweet spot” protest, which is painful enough to generate a response but not painful enough so that an aggressive response is universally applauded. Since non-epsilon protests serve primarily to signal the leaders’ type, they will occur either when protesters have private information about the leader’s type or when the distribution of voters’ preferences are convex in a way that leads the revelation of uncertainty to increase the probability of regime change. The requirements needed for rational civil disobedience seem not to hold in many world settings, and so we explore ways in which bounded rationality by protesters, voters, and incumbent leaders can also explain civil disobedience.