Abstract: Runoff varies with land use types, which in turn affects eco-sustainable development of river basins. In this study, SWAT (Soil and Water Assessment Tool) was used to simulate monthly runoff in relation to land use in Xiangjiang River Basin (XRB). Observed runoff data were collected from the Xiangtan, Zhuzhou, Hengshan, Hengyang and Guiyang monitoring stations. A total of seven model parameters were calibrated with data for 1998~2002 and validated for 2003~2007. The coefficient of determination of linear regression (R²) of the observed and simulated monthly runoff and the corresponding Nash-Sutcliffe Index (NSI) were used to evaluate the model performance. With the exception of Guiyang station, all R² and NSI values were higher than 0.82, with some as high as 0.92. The results showed that SWAT reliably simulated runoff in XRB. Given the high model performance, three scenarios (based on the 2006~2020 comprehensive land use planning in Hunan Province) were used to study the impact of land use type on runoff in XRB. Compared with the existing land use conditions, predicted runoff dropped by 1.28 mm under scenario 1 (where 165.40 km² of agricultural land was changed into forest land and 793.91 km² of agricultural land changed to grassland). Under scenario 2 (where 8 173.96 km² of forest land was reclaimed for agriculture in the upstream basin and 337.56 km² of agricultural land in downstream basin put under urban developed), predicted runoff rose by 15.61 mm. In scenario 3 (where 500.02 km² of unused arable land was reclaimed for agriculture), predicted runoff increased by 1.16 mm. The scenario simulation suggested that runoff dropped under increasing forest land and grassland areas and decreasing paddy field and urban areas. Hydrological effects, economic benefits and human activities were identified as critical factors of runoff and water resources availability in the basin. It was concluded that these factors should be fully taken into account in land use planning and development in XRB.