For automatic train operation (ATO) system, the main object of ATO controller is to control the moving of the train as close as possible to the target curve, so that the demands for time, safety, travelling comfort and energy saving can be met. In this paper, a model predictive control (MPC) strategy is introduced to control the train system due to its good performance and ability to handle system with time-delay and constraints. Model parameters can be identified by measured data. The proposed MPC algorithm with corresponding switching strategy is designed. The simulation results illustrate the feasibility and effectiveness of the proposed control scheme.