Pedestrian evacuation is actually a process of behavioral evolution. Interaction behaviors between pedestrians affect not only the evolution of their cooperation strategy, but also their evacuation paths-scheduling and dynamics features. The existence of interaction behaviors and cooperation evolution is therefore critical for pedestrian evacuation. To address this issue, an extended cellular automaton (CA) evacuation model considering the effects of interaction behaviors and cooperation evolution is proposed here. The infl uence mechanism of the environment factor and interaction behaviors between neighbors on the decision-making of one pedestrian to path scheduling is focused. Average payoffs interacting with neighbors are used to represent the competitive ability of one pedestrian, aiming to solve the confl icts when more than one pedestrian competes for the same position based on a new method. Infl uences of interaction behaviors, the panic degree and the confl ict cost on the evacuation dynamics and cooperation evolution of pedestrians are discussed. Simulation results of the room evacuation show that the interaction behaviors between pedestrians to a certain extent are beneficial to the evacuation efficiency and the formation of cooperation behaviors as well. The increase of confl ict cost prolongs the evacuation time. Panic emotions of pedestrians are bad for cooperation behaviors of the crowd and have complex effects on evacuation time. A new self-organization effect is also presented.