Future runoff prediction is linked to a prior knowledge of future probable precipitation amounts. Hydrological models are a way that can enable transforming observed precipitations into stream flow. In this work a GIS based hydrologic model, SWAT is used in an Upper Narmada River basin at Patan in Jabalpur (M.P). The importance of this thesis comes due to Upper Narmada River is a river in which every year the discharge amount varies according to the weather condition. Some of the months the river is flooded and in some of the years the river is in drought condition. Due to this reason it will be very useful if we can roughly predict the runoff amount. Depending upon the river discharge amount we can decide whether there will be flood or drought condition in the watershed.The catchment area for this watershed has an approximate drainage area of 3950 km2, which is divided into 97 sub basins and 907 HRUs.SWAT model is developed to evaluate different parameters of the water resources like rainfall runoff modelling, sedimentation, eco-hydrological behavior of the watershed. SWAT tool with QGIS software to simulate water runoff for PATAN Watershed in Narmada river basin. It is concluded that the value of R2 can be considered reasonably satisfactory for estimating a watershed runoff. The value of R2 is 0.715.The results from the SUFI-2 calibration process indicated a good performance with a PBIAS value of 12.6% and R2 value was 0.83 indicating a good correlation between observed and simulated values. For the validation these value of R2 and PBIAS are 0.79 and 10.6 % respectively. Results showed that the annual mean flow discharge for the period 1996 to 2005 was 612.52 m3/s which agreed with the model result which shows that the average annual basin simulation values was 524.92 m3/s for the same period. While result simulation for the period 2016 to 2050 gave 414.70 m3/s which mean there is a reduction in mean flow discharge between two periods about 20.99%.