Estimating the impacts of climate change on groundwater represents one of the most difficult challenges faced by water resources specialists. One difficulty is that simplifying the representation of the hydrological system often leads to discrepancies in projections. This study provides an improved methodology for estimation of the impacts of climate change on surface and groundwater reserves, where a physically-based surface–subsurface flow model is combined with advanced climate change scenarios for Hamadan-Bahar basin (2459 km 2 ), Iran. Coupled surface–subsurface flow is simulated with the HydroGeoSphere. The simultaneous solution of surface and subsurface flow equations in HydroGeoSphere, as well as the internal calculation of the actual evapotrairation as a function of soil moisture at each node of the defined evaporative zone, improved the representation of interdependent processes like recharge, which is crucial in the context of climate change. More simple models or externally coupled models do not provide the same level of realism. Climate change simulations were obtained from HADCM3 general circulation model (GCM) scenarios assuming the SRES A2, A1B, B1 emission scenarios. These GCM scenarios were downscaled using LARS-WG stochastic downscaling model. Analyzing the observed weather data showed that distribution of observed daily precipitation is not normal and minimum and maximum daily temperature have autocorrelation. Therefore nonparametric methods were used to compare the observed and simulated data by LARS-WG. Comparison of simulated and observed data and uncertainty analysis of downscaled data verified LARS-WG ability in weather data generation. In calibration step of HydroGeoSphere model, in order to run the model in transient mode and get its responses to daily tensions, the model was first run in steady state in order to apply its results as the initial conditions for transient mode using last 20 years average precipitation data and aquifer discharge. The model was run in transient mode during 1992-2005 period and model parameters were calibrated. In validation step, the model was run during 2006-2010 period and the performance of the HGS model was verified. The combined use of an integrated surface–subsurface modeling approach, a spatial representation of the evapotrairation processes and sophisticated climate change scenarios improves the model realism and projections of climate change impacts on water resources. According to climate change scenarios predictions, except A2 emission scenario at 2071-2100 period which the precipitation is less than control period, future is hotter and wetter than now. Maximum increase in mean annual minimum and maximum daily temperature occurred in A2 emission scenario (4.75 and 4.63 ?C) and minimum increase in mean annual minimum and maximum daily temperature occurred in B1 emission scenario (2.81 and 2.69 ?C). Maximum increase in mean annual daily precipitation occurred in B1 emission scenario in 2011-2040 period (19%) and maximum decrease in mean annual daily precipitation occurred in A2 emission scenario in 2071-2100 period (8%). Integrated flow simulations showed that significant decreases are expected in the groundwater levels by 2100. Maximum and minimum decrease in groundwater level happened in A2 (9 m) and B1 (7 m) emission scenarios, respectively. Total water mass balance didn’t change significantly in comparison with control period. Comparison between groundwater level decrease under different emission scenarios and aquifer discharges showed that severe groundwater withdrawal is the major problem in the study region. Keywords : Climate change, Downscaling, HydroGeoSphere, LARS-WG, GCM.