With the increasing operation of hydropower stations, a refined generation dispatch model becomes more urgently needed for the precise control of cascade hydropower stations under deterministic runoff, such as rules extraction by simulation operation under long series runoff, optimal dispatch level evaluation of historical operation process, and etc. However, most of the studies on refined generation dispatch model focused on the refined description of transform relationships between hydropower and electric energy or hydraulic connection between cascade hydropower stations. The time scale of dispatch model, which is another important factor affecting the accuracy of dispatch model, was short of attention. The time scale of dispatch model affects theaccuracy of dispatch model in a complex way. This paper attempted to discover the influence between time scale and the accuracy of dispatch model by both theoretical analysis and case study. The control variable method and progressive analytical method were used to uncover the key factors which affect the accuracy of dispatch model and to analyze its influence rule qualitatively. The case study of cascade hydropower stations downstream of the Jinsha River has been conducted to analyze influence between time scale and the accuracy of dispatch model by these factors. With modeling analysis in different inflow conditions and water level control strategies, we have uncovered key factors which affect the accuracy of dispatch model quantitatively. Then we proposed a multiple time-scale generation dispatch model for accurate description of the dispatch problem. The simulation results show that the theoretic results proposed in this paper is sound. Moreover, the proposed model has a better performance on both accuracy and efficiency compared with other models, which means that this model can be used for historical operation evaluation and dispatch rules extraction of cascade hydropower stations downstream of the Jinsha River.