Agents face strategic uncertainty in a regime change game that is akin to debt rollover. A mass of agents sequentially decide whether to attack a regime or not, but they do not observe the past actions of other agents. A regime is viable if it can succeed in the absence of any attack. A principal wants to dissuade the agents from attacking a viable regime. We show that if the principal repeatedly runs a "viability test" with sufficient frequency, then the risk that agents may attack a viable regime unravels from the end.