This study assesses the effect of potential future climate change on the inflow of agricultural reservoir using the SLURP (Semi-distributed Land Use-based Runoff Process), a physically based semi-distributed hydrological model. Two SRES (Special Report on Emissions Scenarios) climate change scenarios (A2 and B2) results of CCCma CGCM2 by the IPCC (Intergovernmental Panel on Climate Change) were applied and future land uses were predicted by CA (Cellular Automata) - Markov technique using the land use data of past Landsat images. The SLURP model was calibrated using daily streamflow records of 1998, 1999 and 2006 and validated using 2002 and 2004 streamflow data in the upstream watershed of Gongdo gauging station of Anseong-cheon including Geumgwang and Gosam agricultural reservoir. The calibration and validation results showed that the model was able to simulate the daily streamflow well, with a coefficient of determination and Nash-Sutcliffe efficiency ranging from 0.8 to 0.9 and 0.5 to 0.8, respectively, for both the calibration and validation periods. Future vegetation cover information in the model was also considered by the linear regression between monthly NDVI (Normalized Difference Vegetation Index) from NOAA/AVHRR images and monthly mean temperature using eight years (1998 - 2005) data. The results show that the 2030, 2060 and 2090 inflow of Geumgwang and Gosam agricultural reservoir decreased 49.6 %, 45.6 % and 51.8 %, and 55.8%, 52.6% and 60.6% comparing with the 2004 inflow, respectively. In addition, the streamflow of Gongdo gauging station watershed decreased 53.0%, 48.8% and 57.2%, respectively.