Land use change and climate variability are the main drivers of watershed hydrological processes. The main objective of this study was to assess the impact of land use change and climate variability on hydrology of the Mara River Basin in East Africa. Land use maps generated from satellite images were analyzed using the intensity analysis approach to determine the patterns, dynamics and intensity of land use change. Changes in measured streamflow caused separately by land use change and climate variability were separated using the catchment water-energy budget based approach of Budyko framework. The information on past impact of climate variability on streamflow was used to develop a runoff sensitivity equation which was then used to predict the future impact of climate change on streamflow. Finally, the impact of agroforestry on watershed water balance was predicted using SWAT (Soil and Water Assessment Tool) model. Deforestation and expansion of agriculture were found to be dominant and intensive land use changes in the watershed. The deforestation was attributed to illegal encroachment and excision of the forest reserve. The deforested land was mainly converted to small scale agriculture particularly in the headwaters of the watershed. There was intensive conversion of rangeland to largescale mechanized agriculture which accelerated with change of land tenure (privatization). The watershed has a very dynamic land use change as depicted by swap change (simultaneous equal loss and gains of a particular land use/cover) which accounted for more than half of the overall change. This implies that reporting only net change in land use (of MRB) underestimates the total land use change. The results show that streamflow of Nyangores River (a headwater tributary of the Mara River) significantly increased over the last 50 years. Land use change (particularly deforestation) contributed 97.5% of change in streamflow while the rest of the change (2.5%) was caused by climate variability. It was predicted that climate change would cause a moderate 15% increase in streamflow in the next 50 years. SWAT model simulations suggested that implementation of agroforestry in the watershed would reduce surface runoff, mainly due expected improvement of soil infiltration. Baseflow and total water yield would also decrease while evapotranspiration would increase. The changes in baseflow (reduction) and evapotranspiration (increase) were attributed to increased water extraction from the soil and groundwater by trees in agroforestry systems. The impact of agroforestry on water balance (surface runoff, baseflow, water yield and evapotranspiration) was proportional to increase in size of the watershed simulated with agroforestry. Modelling results also suggested that climate variability within the watershed has a profound effect on the change of water balance caused by implementation of agroforestry. It is recommended that authorities should pay more attention to land use change as the main driver of change in watershed hydrology of the basin. More effort should be focused on prevention of further deforestation and agroforestry may be considered as a practical management strategy to reverse/reduce degradation on the deforested parts of the watershed currently under intensive cultivation.