This study contributes to the discussions on the future of the Amazon rainforest under a projected warmer-drier climate from the perspectives of land hydrology. Using IPCC HadGEM2-ES simulations of the present and future Amazon climate to drive a land hydrology model that accounts for groundwater constraint on land drainage, we assess potential hydrologic changes in soil water, evapotranspiration (ET), water table depth, and river discharge, assuming unchanged vegetation. We ask: how will ET regimes shift at the end of the 21st century, and will the groundwater help buffer the anticipated water stress in some places-times? We conducted four 10 yr model simulations, at the end of 20th and 21st century, with and without the groundwater. Our model results suggest that, first, over the western and central Amazon, ET will increase due to increased potential evapotranspiration (PET) with warmer temperatures, despite a decrease in soil water; that is, ET will remain PET or atmospheric demand-limited. Second, in the eastern Amazon dry season, ET will decrease in response to decreasing soil water, despite increasing PET demand; that is, ET in these regions-seasons will remain or become more soil water or supply-limited. Third, the area of water-limited regions will likely expand in the eastern Amazonia, with the dry season, as indicated by soil water store, even drier and longer. Fourth, river discharge will be significantly reduced over the entire Amazon but particularly so in the southeastern Amazon. By contrasting model results with and without the groundwater, we found that the slow soil drainage constrained by shallow groundwater can buffer soil water stress, particularly in southeastern Amazon dry season. Our model suggests that, if groundwater buffering effect is accounted for, the future Amazon water stress may be less than that projected by most climate models.