Temperature anomaly series at 160 stations of China in winter(December,January and February) from 1951 to 2010 are processed by the station homogeneity adjustment scheme.The adjusted series for dealing with the variance heterogeneity distribution caused by station network heterogeneity are more effective.EOF(empirical orthogonal function) analysis method is used to study the original and adjusted temperature anomaly series.Results show that:1)There is better organization for high absolute value areas of the first two typical fields of the adjusted data than those of the original data because the high absolute value areas of the typical fields locate mainly in the three north region(Northwest China,North China and Northeast China) and the Tibetan Plateau,which are consistent with the high value areas of mean square deviation field.2)The evolution features of main mode time coefficients of before and after adjustment are similar r_h≥0.886,h=1,2),but the variance contribution of linear and interdecadal components of the first two time coefficients for the adjusted data much more focus on the first mode.3)The linear components of the first typical fields and their time coefficients for the adjusted data can display the main strong warming areas over China in winter during the past 60 years,and their interdecadal components are more consistent with the evident interdecadal variation areas over China than those of the original.