To improve the efficiency of adjacent metro trains entering and leaving a station, we propose an improved genetic algorithm (IGA) which introduces a combinational mutation strategy to the classical genetic algorithm. Based on this algorithm, the running processes of adjacent metro trains within stations are optimized. The process is primarily divided into three stages: posterior trains entering the station, posterior trains stopping at the station, and previous trains leaving the station. These stages are principally influenced by four factors: the acceleration and initial speed of the posterior train entering the station; the time when the previous train leaves the station; and the acceleration of the previous train leaving the station. Moreover, there are certain coupling features and relationships among these factors. How to search for the optimal values of these factors is the issue to be discussed in this paper. Experiment results show that we can obtain an optimal solution in the space established by a suitable combination of values for these factors, and that the IGA is potentially useful for optimization design. © 2013 Institute of Electrical Engineers of Japan. Published by John Wiley & Sons, Inc.