This study analyzed the ERA-Interim data from the European Centre for Medium-Range Weather Forecasts and NECP reanalysis data from the National Oceanic and Atmospheric Administration Earth System Research Laboratory in the Antarctic region to compare with the radiosonde data derived from the Antarctic 18 stations, assessed the applicability of these two reanalysis data sets in the upper atmosphere.The results demonstrated that the reanalysis data sequentially drift away from the radiosonde data steadily in four meteorological elements with the increase in height over the Antarctic.The biases in geopotential height and temperature in the two reanalysis data are relatively small,but the differences of wind direction between the sounding data and the two sets of reanalysis data are quite huge.The differences in wind speed between the sounding data and the two sets of reanalysis data are quite large at 300 hPa.In the seasonal distributions of these biases,the differences of the geopotential height and temperature between the sounding data and the two sets of reanalysis data are huge in the austral spring but are relatively small in autumn,summer,and winter.The difference in the wind speed between the sounding data and the two sets of reanalysis data is relatively small in summer.The difference of the wind direction between the sounding data and the two sets of reanalysis data is relatively large but has no significant seasonal dependency.Nevertheless,both reanalysis data sets have large biases,ERA-Interim performs better than NCEP as a whole.The results suggest that using the data from these two reanalysis data sets as initial and boundary conditions to drive regional atmospheric models can lead to large model biases.Thus,more intensive radiosonde observations are necessitated for improving the qualities of reanalysis data,along with further improvements in the data assimilation and model system.