Abstract: At present, the flood evacuation path in emergency plan for reservoir dam safety management is based on short time and short distance principles, which are usually adopting static paths and cannot dynamically follow the dam-break flood routing. The BREACH-MIKE21 coupling model was used to simulate the dam-break flood routing to obtain the dam-break flood submergement information, and combine the road network model to build an evironmental scenario model. The ant colony algorithm was used to optimize the flood evacuation path, and through adding backtracking, introduce the maximum and minimum ant colony systems, and improve the heuristic function to better the optimization ability and convergence speed. By analyzing flood submergence elements frame by frame, the dam-break flood information was integrated into the algorithm, and the optimal flood evacuation path was finally obtained. The simulation results show that: (1) Under the same inundation scenario, compared with the basic ant colony algorithm, the average running time of the improved ant colony algorithm is increased by 1.6‰, the accuracy is increased by 32%, the average convergence times are reduced by 10.13 times, the convergence speed is faster and the optimization ability is stronger. (2) The optimal dynamic flood avoidance path obtained by the improved ant colony algorithm based on the evolution of dam-break flood is safer than the static path. The results of this study suit the requirements of emergency plan for reservoir dam safety management in the dynamic process of flood avoidance transfer path, and can improve the efficiency of emergency transfer and reduce the impact of dam-break floods.