In the emergency evacuation of pedestrians in subway stations, most people do not know the correct evacuation routes and methods, and tend to follow others blindly, resulting in herding behavior. To study the process of pedestrian evacuation with herding behavior in subway stations, a cellular automata pedestrian evacuation model is established to simulate the pedestrian evacuation with herding behavior in subway stations, and the effects of the weight coefficients of the parameters in the model on the overall evacuation time and the distribution of people’s positions are investigated to quantify the effects of herding on the efficiency of the pedestrian evacuation and the movement behavior. The simulation results show that moderate (kC < 40) crowd behavior can play a beneficial role in reducing reaction time and guiding the direction of evacuation; on the contrary, excessive crowd behavior (kC > 40) will lead to an increase in evacuation time and a decrease in evacuation efficiency. The conclusions presented in this paper can help to improve the efficiency of pedestrian evacuation in emergencies and provide a theoretical basis and practical guidance for the management of safety evacuation in subway stations.