In order to improve the efficiency of emergency evacuation of subway passengers, the concept of ‘guide partition’ is put forward, and an intelligent algorithm of guide partition for emergency evacuation of subway stations is proposed according to the theory of allocation problem in operational research. First an evacuation partition optimization algorithm based on the shortest path principle and based on exit capacity for non‐standard allocation problem is proposed, according to the solution of service facility allocation and site selection problem in operational research. Secondly, the passengers evacuation area is divided to partitioned evacuation zones. Combined with the advantages of cellular automata model, the evacuation simulation software of Building Exodus is used for simulation and comparison of the advantages of the proposed intelligent algorithm. Finally, the evacuation partition intelligent algorithm considering the exit capacity is applied to the Military Museum Station of Beijing subway to obtain the numbers and distribution of guiders on each floor. The results show that the proposed algorithm can help the subway station staff to better determine the number, location and influence range of each guider, and effectively improve the utilization rate of evacuation exits, reduce evacuation time and improve evacuation efficiency..