Abstract: Resigning in protest is the strongest means by which government officials can register disapproval of a leader's actions. Doing so may facilitate electoral accountability, or may simply result in less principled agents filling the void. I develop a formal model that examines the informational impact of principled resignation, and the cross-cutting incentives that officials face for bringing their criticisms to the public versus staying and improving governance from within. Two classes of equilibria are examined. First, in a "protest" equilibrium, agents are more likely to resign under bad leaders than good leaders; the direct effect of removing themselves as a safeguard against bad policy, and the indirect effect of signaling disapproval, work in conjunction to enhance electoral accountability. Second, in a "public servant" equilibrium, agents are more likely to stay and serve under bad leaders than good leaders; electoral accountability in this equilibrium is weaker, but voter welfare may be improved. If the agent has any impact on policy outcomes, her resignation---even when driven purely by career or financial motives---cannot be apolitical.