In order to explore the impact of hydrological variation and external diversion of water supply on power generation and ecology of hydropower stations, taking the downstream water-reducing reach of Guanting Reservoir as an example, based on the calculation of ecological flow under hydrological variation, the ecological assurance degree of the water-reducing reach is defined. A multi-objective optimal operation model for reservoir power generation and ecological guarantee degree is established, and the non-dominated sorting genetic algorithm (NSGA-II) is used to solve multiple scenarios. The results show that the inflow runoff of Guanting Reservoir changed in 1971 and 1983 respectively, and the power generation and average ecological assurance degree decreased significantly after the hydrological variation. The multi-objective relationship between hydropower generation and ecology is significant, and the reduction of 37.4% in power generation can increase 67.8% in ecological benefits. The power generation and the average ecological guarantee degree are significantly improved after the recharge of the external water source, especially in the dry year, which increased by 56.9% and 40.3% respectively.