Global terrorism and local rebellion are observed rather frequently; solutions appear to be rather sparse. A common strategy adopted by governments is to attempt to deter potential rebels from engaging in acts of violence, often by responding to attacks with violent reprisals directed at the populace from which the rebels are recruited. The logic supporting such a strategy follows from a theory of deterrence that we believe is underdeveloped. This theory focuses heavily on the credibility of the deterrent threat and on ensuring that the cost imposed by retaliation is high. This ignores the flip side of the deterrent threat—that is, the promise that if one does not misbehave (by engaging in acts of rebellion), one will not suffer retaliatory punishment. Policy designed to ensure that punishment is swift and severe (and perhaps disproportional) can undermine deterrence and actually encourage more potential rebels to become active. Using a game-theoretic model, we show that successful deterrence requires a strategy in which retaliation is proportionate and directed only at the guilty, as well as being certain. Moreover, potential rebels must believe that the innocent will have attractive opportunities outside of joining the insurgency as well as that the innocent will not be punished. We provide empirical illustrations of our thesis from participants in the Palestinian uprising, including several who have taken up arms against Israel.