The availability and geographic distribution of fresh water resources may undergo significant changes in response to global environmental change. In this study, we examine the water balance of the Amazon Basin using a modified version of the LSX land surface model <Pollard and Thompson, 1995; Thompson and Pollard, 1995a, b> which includes a representation of land surface processes, canopy physiology (stomatal conductance, transpiration, and photosynthesis), and continental-scale hydrological routing. The model operates on a 0.5 ¿C by 0.5 ¿C grid and is forced with observed long-term climatological data. As an initial application of the model, we examine the seasonal variability of water balance within the Amazon Basin. The simulation is evaluated by comparing (1) simulated evapotranspiration with observations for different vegetation cover types and (2) simulated river discharge against the long-term records of 56 fluviometric stations spread throughout the basin. The model results show that evapotranspiration is strongly dependent on the vegetation cover, especially during the rainy season. Overall, we find good agreement between the simulated and the observed water balance: for most of the fluviometric stations the error is less than 25%. In addition, we perform a model sensitivity study to determine the role of changes in vegetation cover on the water balance, without considering feedbacks on climate. When forests, woodlands, and savannas are replaced with grasslands, annual average evapotranspiration decreases by ~0.5 mm d-1 (~12%), which is comparable to observations. Finally, we perform a model sensitivity study in order to assess the potential physiological effects of increased CO2 on stomatal (canopy) conductance and, as a consequence, on the water balance of the Amazon Basin, again without considering feedbacks on the atmosphere. The model results suggest that doubling atmospheric CO2 concentrations (from 325 to 650 ppmv) would decrease the canopy conductance by 20 to 35% (depending on the vegetation type) and would decrease evapotranspiration by ~4% throughout the region. As a consequence, annual river discharge increases by between 3% and 16.5%, depending on the position within the basin. At the mouth of the Amazon and Tocantins Rivers, annual discharge increases by 5 and 9%, respectively. ¿ 1997 American Geophysical Union