Even in authoritarian states, dissent is not always repressed and repression occasionally seems to backfire and is followed by an increase in dissent. I propose a simple explanation: When a population is uncertain whether it is facing a government whose interest in conducting a certain policy is aligned or misaligned with its own, repression may signal misaligned interest as the government then has more to lose from policy change, thus having more to gain from repression. The population then expects to gain more from dissent. In a game between a government and a population, I show that also a government whose interest is misaligned with the population’s may not always repress, simply because the signalling effect of not repressing decreases the expected gain of dissent for the population. This holds even when repression quells dissent completely. When repression cannot hinder dissent, but instead increases the cost of dissent to the population and reduces its ability to induce policy change, at intermediate levels of costs vs. gains for the government, signalling misalignment means that only NE in which repression is always followed by dissent are PBNE. The game highlights the importance of factors such as institutions that influence the cost of repression. An extension shows that a government who is uncertain about the population’s valuation of policy change is willing to pay for bulk surveillance.