The abilities of the National Aeronautic and Space Administration Goddard Institute for Space Studies (GISS) and METEO-FRANCE Arp¿ge General Circulation Models (GCMs) of the atmosphere to simulate surface climate variables that have direct and indirect impacts on the mass balance of the Antarctic ice sheet are evaluated. Deficiencies in reproducing the surface temperature are identified. Some of the flaws are associated with an incorrect specification of the Antarctic continent real topography, a problem which can be partially corrected by an adequate processing of model output. Shortcomings can also be consequent to the use of inappropriate climate variable formulations and prescribed boundary conditions. A coarse-resolution version of the GISS GCM is used to demonstrate how model results can be improved if snow albedo and sea ice coverage are more adequately set. A barrier to thoroughly analyzing and validating model results is the lack of numerous and reliable enough observational data. This point is particularly critical for accumulation, a first-order term in the ice sheet mass balance. Although accumulation is very different in the GISS and Arp¿ge models, uncertainties about the available observational references preclude firmly asserting that one model is better than the other. Finally, typical GCM resolution remains an essential limitation to improving model performances and promoting the effective use of model results for the Antarctic region. ¿ American Geophysical Union 1994