If checks and balances are aimed at protecting citizens from government's abuse of power, why do they sometimes vote to weaken them? We address this question in a laboratory experiment in which subjects are asked to choose between two decision rules that resemble real world institutional environments with and without checks and balances. In our theoretical framework, voters may prefer an unchecked executive if that enables a reform that, with checks and balances, is blocked by the legislature. Consistent with our predictions, we find that subjects are more likely to weaken checks and balances when there is political gridlock, i.e. when the executive proposes a reform and the legislature blocks it. However, subjects in the lab weaken the controls not only when the reform is beneficial, which is the predicted result, but also when it is harmful. In this sense, subjects exhibit an overreaction to political gridlock, providing excessive special powers to the executive. This overreaction entails a loss in the voters’ payoffs.