On the premise of not affecting the existed interest patterns in the context of flood control and water supply, the major scientific issue before us is how to improve the supply guarantee rate of ecological water for northern China water scarce regions, through improving reservoir operation rules and maximizing efficiency in the joint operation for multi-reservoir systems. A multi-reservoir ecological operation model is developed based on the subdivision of reservoir storage capacities. The model treats the river ecosystem as an independent water user, and formulates a scheduling scheme for partitioning multi-reservoir storage capacities for multiple water users. The effect of the scheduling scheme on the ecology and power generation benefits is considered through the establishment of a simulation module for the river basin water resources system. The module is used in combination with an objective function for the development of multi-reservoir ecological operation model. The purpose of the objective function is to maximize the supply guarantee rates while ensuring a balance as possible among all water users. The multi-reservoir ecological operation model is applied to the Luanhe River system of Haihe River basin. As the result of the application, the scheduling scheme is proposed for the joint operation of three reservoirs including Panjiakou, Daheiting and Taolinkou under the existed water supply conditions. The operational chart is provided on subdivision of reservoir storage capacities. The effect of the scheduling scheme is evaluated on the downstream ecological system and power generation benefits. The proposed model is proven to be effective by overall modeling results. Comparing to the current scheduling scheme, the new scheduling scheme can significantly increase the supply guarantee rate of ecological water from 46.9% to 93.8%, while not affecting the existed interest patterns in the context of flood control and water supply. The index for measuring eco-hydrological changes is also improved significantly.