I propose a simple simulation procedure for large games with multiple equilibria. The simulation procedure is based on a best-response dynamic. The implied equilibrium selection mechanism is intuitive: more stable equilibria are selected with higher probability