SWAT model has been widely used in simulating the effects of land use change on hydrology, but its application to assessment of the impacts of major shifts in land use needs further enhancement. The accuracy of this model is reduced by homogenizing the slope and spatial difference of land use. SWAT model has been modified in this study considering the CN value of land use type (especially the forest) to different slopes and the forest hydrological adjusting function, to improve its sensitivity to changes in land use and its runoff simulation. The area of Lao Guanhe watershed was selected in a case study by dividing each of its sub-basins into four slope grades, and the CN value was calculated for each land use type in different slope grades; and the SWAT model was modified to achieve the forest hydrological adjusting function in each slope grade through improving the methods for estimation of surface runoff. The results show that a significant improvement was achieved in monthly and daily hydrological simulations in comparison to the original SWAT simulations for both model calibration and validation periods, especially in rainy season. The modified SWAT model considering the forest hydrological adjusting function can provide better response to climate change and land use change.