A new cellular automata traffic model based on Revised S-NFS model is established to consider a mixed flow system in which the maximal velocity of the agents is distributed, as is the case in a real traffic flow fields composed of compact vehicles, trucks, and buses. These vehicles are assigned on of two different strategies: cooperator (C), who remains in his original lane, and defector (D), who undertakes lane-changing to maximize his own payoff, i.e., average velocity. In a systematic series of multi-agent simulations, we quantitatively compare flow characteristics in the default system (where maximal velocity is constant across all agents), mixed traffic flow systems (which permit a distribution of maximal velocities), and correlated–mixed traffic flows, in which an agent with a higher maximal velocity tends to have a D strategy whereas one with a lower maximal velocity tends to have a C strategy. We discuss what kind of game class is underlying in each traffic flow system. Furthermore, we quantitatively study the social efficiency deficit, an index of dilemma extent, for each of the flow systems.