A comprehensive hydrological study of large watersheds in Africa e.g. the Congo basin and the Nile basin has not been vigorously pursued for various reasons. One of the major reasons is the lack of adequate modeling tools that would not be very demanding in terms of input data needs and yet inclusive enough to cover such wide extents (over 3 million square kilometers for the Congo basin). Using a coupled run of the Community Atmospheric model (CAM3) and Community Land Model (CLM3) components of the Community Climate System of Models (CCSM), this study looks into the spatial and temporal variation of precipitation and river runoff in the Congo basin in the light of increasing trends in deforestation of the tropical forests. The effect of deforestation on precipitation and runoff is investigated by changing the land cover-type from the current configuration of broadleaf evergreen/deciduous, non-Artic grass and corn to a mostly grass type of vegetation. Discharge simulation for the river Congo is centered at the point of entrance to the Atlantic Ocean. Although the CLM3 does not presently simulate the observed river runoff to within at least one standard deviation it gives an opportunity to iteratively improve on the land surface parameterization with a possibility of future accurate prediction of mean monthly river runoffs under varying climate scenarios and land use practices. When forced with the National Center for Environment and Prediction (NCEP) re-analysis data the CLM3 runoff simulation results are relatively more stable and much closer to the observed. An improved CLM3 when coupled to CAM3 or other Global Climate Models is definitely a better tool for investigative studies on the regional hydrological cycle in comparison to the traditional methods. There was a slight reduction in rainfall in the first experiment which mimicked a severe form of deforestation and a slight increase in rainfall following low level of deforestation. These changes in rainfall were however statistically insignificant when compared to the control simulation. There was notable heterogeneity in the spatial distribution of the changes in rainfall following deforestation.