This study develops an optimal reservoir management model based on each ten-day operation periods for Shihmen Reservoir. The model tries to not only meet the human water requirement, but also to consider water quality and ecological issues. In Taiwan, lots of infrastructures have been built in almost every river to effectively use water resources and provide disaster defense. It results in downstream flow regulation or violent flow fluctuation that causes impacts on aquatic ecosystem biodiversity. Decision makers should take flow variability into account in water resources management policy. Human demands, water quality, and ecological issues are considered in the optimal reservoir management model. Human demands objective is based on water rights of different users and M-5 Rule Curve of Shihmen Reservoir. Ecological objective is based on consideration of Environmental flow components (Richter et al. 2007). Water quality model uses an artificial neural network model to simulate downstream water quality by using streamflow, rainfall, upstream water quality, and measure time as inputs. Then this model is combined with human demands and environmental flow components considerations to establish an optimal reservoir management model by using genetic algorithm. The optimal reservoir operation model considers human demands, water quality and flow regime to benefit to both human society and aquatic ecosystems.