Abstract: In order to analyze the evacuation behavior of the crowd in the emergency situation from the perspective of emotion,existing works on emotional contagion are reviewed and behavior characteristics of the crowd in emergency situations are summarized.A multi-agent emotional contagion model is proposed,in which individuals in the crowd are represented by agents.Its main frame is divided into perception layer,emotion layer,infection layer,behavior layer and action layer.Three conditions for the generation of emotional contagion and three rules for emotional contagion are summarized,algorithm for emotional contagion is proposed and individual personality and inter-individual distance factors are considered.Emotional intensity and crowd density are used to calculate the evacuation rate of the individual.A simulation experiment was compiled in C#language.Evacuation time is basically consistent in the simulation and in the real earthquake case.Compared with the existing emotional contagion model that based on infectious disease ideas,the proposed model can better describe the contagion process from local to global.The experimental results show that the proposed model can deduce crowd’s emotion-driven aggregation behaviors.It could be a visual analysis method for the formulation of emergency evacuation plans.