The physically based Soil Water Assessment Tool (SWAT) model is applied on the Simiyu river catchment at Ndagalu outfall, Tanzania. Input data are spatially distributed data set such as topographical, land cover and soil data, climatic and river flow time series data. The model was calibrated for the period 1970-1974 based on the availability of coinciding climatic data and validated in the period 1976-1983. The calibration involved fine-tuning the land coversoil parameters to match the observed discharge at Ndagalu flow gauging station. The weather generator tool of the SWAT was used to fill in the missing climatic data and enabled flow simulation in the periods with missing data. Several model fit techniques and water balance matching techniques were used to assess model performance. Results and analysis was presented to show model performances in calibration and validation periods. During model set-up it was found that the land use (especially vegetation types) classification was a sensitive issue for the river flow estimation. The study identified optimum parameters for better discharge simulations. However, since some model parameters were obtained manually by fine-tuning, in future study the more physical basis of the parameters need to be considered. Also, the analysis of observed and estimated hydrographs showed that the rainfall data used is not representative and more rainfall data is required especially in the headwaters of the catchment so as to have better rainfall representation, which may produce better model fits. Considering the good results of SWAT in this study and comprehensiveness of the model in land surface processes representation, the model is very promising for land and water management studies and expected to give valuable information to land and water resources managers.