Reanalyses are extensively used for studies of Arctic climate variability and change, and are often assumed to be reliable without any validation being attempted. Francis (2002) compared tropospheric winds measured by the Polarbjorn in the eastern Arctic (based on Omega signal tracking) with those analyzed by reanalyses from the National Centers for Environmental Prediction/National Center for Atmospheric Research (NNR) and European Centre for Medium-Range Weather Forecasts (ERA-15). Importantly, these upper level winds were not assimilated into the reanalyses and thus constitute an independent test of the reanalyses fidelity. The reanalysis winds were found to be too westerly and too northerly in relation to the Polarbjorn winds collected during the Coordinated Eastern Arctic Expedition (CEAREX) in 1988/89, especially in the mid and upper troposphere. These findings are confirmed here and the large differences imply serious shortcomings to the reanalyses at least in this part of the Arctic. The comparison has been extended to the new ECMWF reanalysis (ERA-40) and essentially the same results are found. Because the CEAREX study region is adjacent to a number of fixed rawinsonde sites (e.g., on Svalbard), comparisons are made between the reanalysis mid and upper tropospheric winds with those from contemporary rawinsondes in the HARA archive and from the Polarbjorn. Large differences are found between the Polarbjorn winds and those from the other two data sources. The cause of the discrepancy is explored and the implications are discussed.