[Purposes] This paper aims to explore the optimization of customized bus routes with multiple parking lots considering the priority of disabled passengers and provide a theoretical basis for the decision-making of customized bus routes. [Methods] A customized bus route optimization model was built with the goal of minimizing the sum of vehicle travel time cost, fixed vehicle dispatching cost, and waiting time cost for disabled passengers, involving elements such as multiple origins and destinations, multiple parking lots, multiple vehicle types, multiple passenger types, and priority for disabled passengers to get on and off the bus. The Benders decomposition algorithm was improved. The multi-plane cut constraint strategy was adopted, and the slack variables were introduced to speed up the algorithm solving efficiency. Finally, the model and algorithm were analyzed through a case. [Findings] Compared with the case without considering the priority of disabled passengers, the waiting time and travel time of disabled passengers after considering their priority are reduced by 30.6% and 1.3% respectively. Compared with the traditional branch-and-bound algorithm, the improved Benders decomposition algorithm can obtain satisfactory solutions while reducing the calculation time by 20.6%. [Conclusions] The model proposed in this paper effectively reduces the waiting time and travel time of disabled passengers, and the improved Benders decomposition algorithm is effective. The results are of important practical significance in solving the travel problem of the disabled.