Abstract: Using the NCEP/NCAR re-analysis data (re-analysis I), the ERA-Interim re-analysis data, and Hadley Center Arctic sea ice concentration data, main characteristics of atmospheric circulation anomalies in the winters of 2011/2012 and 2015/2016 were investigated based on diagnostic analysis and numerical simulations, and the possible causes were revealed. The results indicate that although sea surface temperature anomalies in the tropical Pacific in the two winters corresponded to a weak La Nia and a strong El Nio event respectively and were distinctly different, the Siberian high was very strong in both winters, ranking the highest and the 5th highest since 1979, respectively. Autumn Arctic sea ice loss is a major reason that causes strong Siberian high in both winters. More important, anticyclonic surface winds in the preceding summer season over the Arctic Ocean and higher than normal air temperatures in the mid- and low-troposphere aloft strengthen the negative feedback of Arctic sea ice loss on the winter atmospheric variability, which further enhances the Siberian high and thereby favors the occurrence of extreme cold events in East Asia. Apparently the dynamic and thermal dynamic conditions of summer Arctic atmospheric circulation not only influence Arctic sea ice in the summer and autumn, but also modulate the effects of Arctic sea ice loss on winter atmospheric variability. The strong El Niño event in the winter of 2015/2016 couldn't offset effects of Arctic sea ice loss and atmospheric circulation anomalies on the winter atmospheric variability over East Asia.