Axelrod has developed an evolutionary approach to the study of repeated games and applied that approach to the Prisoners' Dilemma. We apply this approach, with some modifications in the treatment of clustering, to a game that has the Prisoners' Dilemma and Chicken as special cases, to analyze how the evolution of cooperation differs in the two games. We find that the main barrier to the evolution of cooperation in Chicken is that cooperation may not always be correctly thought of as socially optimal, but that strong forces do push the players toward socially optimal action. We derive some of the results on mixed populations for any game with pairwise interaction.