In the present paper I introduce the revelation principle of bounded rationality. The principle states that any behavior that a model of bounded rationality exhibits can be implemented as the choice of an agent with preferences and beliefs consistent with Savage-Bayesian rationality. The principle provides a way to unify the two main research programs on the foundations of game theory: the decision-theoretic program and the bounded rationality program. I illustrate how the principle works by providing a solution to the ``El Farol'' problem, a deterministic model of bounded rationality that exhibits stochastic equilibrium behavior. The unification shows that one need not go outside the received theory of Savage-Bayesian decision making to provide adequate foundations for equilibrium analysis.
