This paper focuses on the demand-oriented passenger train scheduling problem for a congested urban rail line, considering uneven spatial and temporal demand distributions. A passenger-train interaction framework is developed to dynamically assign passengers to capacitated trains. A mixed integer nonlinear programming model that combines heterogeneous headways and short turning as an integrated strategy (HH-ST) is proposed with the aim of jointly minimizing passenger waiting time and operational costs, as well as balancing train loads. A two-stage genetic algorithm based on an integer coding approach is proposed to solve this problem. The proposed HH-ST strategy is compared with alternative strategies, namely ST alone, HH alone and regular schedule, through a real-world case study of Shanghai Metro Line 9. The results show that the HH-ST strategy provides a better trade-off between users’ and operators’ cost than other strategies, thus achieving a better match between transport capacity and passenger demand.