Existing deterministic optimization methods suffer from decreased performance and increased risks under conditions of passenger flow uncertainty. To address these issues, a multi-scenario robust optimization method for coordinated optimization of train flow and passenger flow in regional rail transit systems is proposed. Firstly, samples of passenger travel demand, which are sampled from the potential passenger demand distribution, is used as multiple scenarios to characterize the uncertainty and diversity of passenger flow. Secondly, the mean-variance theory is employed as the foundation to establish the robust optimization model and the model’s performance is discussed using variance or integration of deviations as indicator of robustness. Finally, a genetic algorithm is applied to solve the model, and the data from the Chongqing regional rail transit system is used as a case study for validation. Experimental results demonstrate that the proposed robust optimization model outperforms the deterministic model in uncertain conditions, providing better optimization performance.