The interannual variability of East Asian winter monsoon (EAWM) circulation from the Development of a European Multi-Model Ensemble (MME) System for Seasonal to Inter-Annual Prediction (DEMETER)hindcasts was evaluated against observation reanalysis data.We evaluated the DEMETER coupled general circulation models (CGCMs)’ retrospective prediction of the typical EAWM and its associated atmospheric circulation.Results show that the EAWM can be reasonably predicted with statistically significant accuracy,yet the major bias of the hindcast models is the underestimation of the related anomalies.The temporal correlation coefficient (TCC) of the MME-produced EAWM index,defined as the first EOF mode of 850-hPa air temperature within the EAWM domain (20°-60°N,90° 150°E),was 0.595.This coefficient was higher than those of the corresponding individual models (range:0.39-0.51) for the period 1969-2001; this result indicates the advantage of the super-ensemble approach.This study also showed that the ensemble models can reasonably reproduce the major modes and their interannual variabilities for sea level pressure,geopotential height,surface air temperature,and wind fields in Eurasia.Therefore,the prediction of EAWM interannual variability is feasible using multimodel ensemble systems and that they may also reveal the associated mechanisms of the EAWM interannual variability.