Population growth, climate change, institutional change and urbanization etc. would cause the change of land use / land cover, which in turn would further affect the hydrological process (e.g. infiltration, surface runoff, evapotranspiration and lateral flow) and water balance. Currently, because of the difficulties in obtaining the yearly land use / land cover data and the problems in calculating process in hydrological models, the simulation of hydrological process under the impact of land use / land cover change is conducted on one period each time and the impacts of multiple periods (e.g. multiple years) of land use / land cover change cannot be simulated continuously. That is the models could not call land use / land cover data year by year seamlessly. SWAT (Soil and Water Assessment Tools) is a distributed hydrological model. During its simulation period, it could not call land use / land cover data year by year and neglects the land use / land cover change in time scale. This would impact its applications in regions with severe land use / land cover changes (e.g. the Middle Reaches of the Heihe River). Heihe River Basin is the second largest and a typical endorheic river basin in China. The middle reach consumes a major portion of the water in the Heihe River. For the weakness of SWAT model, this research developed a model called LU-SWAT that could simulate the impacts of multiple years of land use / land cover change continuously. Furthermore, we applied SWAT and LU- SWAT to simulating streamflow in the Middle Reaches of the Heihe River, and compared their efficiency. The results showed that the LU- SWAT model, which could call land use / land cover data year by year is more suitable to simulate the streamflow in the Middle Reaches of the Heihe River.