This paper analyzes the equilibrium play in a random matching model with a changing environment. Under myopic decision making, players adopt imitation strategies similar to those observed in evolutionary models with sampling from past play in the population. If the players are patient, equilibrium strategies display elements of experimentation in addition to imitation. If the changes in the environment are infrequent enough, these strategies succeed in coordinating almost all of the players on the dominant action almost all of the time. The myopic rules, on the other hand, result in mis-coordination for a positive fraction of time.