The surface air temperatures observed from stations in Antarctica have been shown to have predominantly positive trends that are as high as 0.5°C decade−1 along the Antarctic Peninsula. To evaluate whether the trends are caused by a local or large-scale phenomenon in the Antarctic region, surface temperatures inferred from infrared satellite data from 1979 to 1998 have been analyzed in combination with data from 21 stations that have long record lengths. The surface temperatures derived from infrared data are coherent spatially and temporally and are shown to agree well with Antarctic station data with a correlation coefficient of 0.98 and a standard deviation of about 3°C. The trend analysis on station data yielded on the average 0.012 ± 0.008°C yr−1 and −0.008 ± 0.025°C yr−1 for the 45- and 20-yr record, respectively. The latter reasonably agrees with the trend of −0.042 ± 0.067°C yr−1 inferred from the satellite 20-yr record. The 20-yr record length is shown to be about the minimum length requ... Abstract The surface air temperatures observed from stations in Antarctica have been shown to have predominantly positive trends that are as high as 0.5°C decade−1 along the Antarctic Peninsula. To evaluate whether the trends are caused by a local or large-scale phenomenon in the Antarctic region, surface temperatures inferred from infrared satellite data from 1979 to 1998 have been analyzed in combination with data from 21 stations that have long record lengths. The surface temperatures derived from infrared data are coherent spatially and temporally and are shown to agree well with Antarctic station data with a correlation coefficient of 0.98 and a standard deviation of about 3°C. The trend analysis on station data yielded on the average 0.012 ± 0.008°C yr−1 and −0.008 ± 0.025°C yr−1 for the 45- and 20-yr record, respectively. The latter reasonably agrees with the trend of −0.042 ± 0.067°C yr−1 inferred from the satellite 20-yr record. The 20-yr record length is shown to be about the minimum length requ...