Land use change in a watershed results in the changing water quality in the receiving waterbody. This causality has been confirmed descriptively but the quantitative relationship was lacked. This study used the observed data in the Feitsui Reservoir watershed to identify the effects of water quality caused by streamflow and land use. Pearson product-moment correlation was used to define the significance and multiple regression analysis was applied to link the relationship. Results show that landuse change affects significantly on suspended solid, total phosphorous, and ammonia-nitrogen, and the less effects on biochemical oxygen demand and dissolved oxygen. Urban land, agricultural land, and grassland have positive significant effects on nutrients and suspended solid, which means tmore urban, agricultural, and grass land in a watershed results in more nutrient and suspend solid concentration in receiving waterbody. On the contrary, forest land has negative effects on these pollutants. A multiple regression formula consisted of several land use was finally built to predict water quality. Combining with land use scenario and the regression equations, the future water quality can be predicted.