Abstract: Using the retrospective forecasts covering 30-year period (1982-2010) provided the precipitation field and geopotential height field data and the forecast products during June to August of 2015 from the second-generation system of Climate Prediction (NCEP CFSv2),NCEP CFSR reanalysis data information,and the observed precipitation data in China,the predictive skills and errors between forecast and observation of China regional precipitation and circulation in summer of 2015 (El Nino developing year) have been evaluated. And the possible factors that affect the model forecast skill level have been analyzed. The results show that: 1) The model predictive skills show obvious monthly variation (in July is the highest,secondly is in August,the lowest is in June),and depend on the length of time in advance,but the overall levels are not very high. 2) CFSv2 can predict the crucial areas of circulation anomalies at 500 h Pa which affect regional precipitation of China in summer very well. The model can advance 5 ~ 9 days accurately predict the spatial mode of next month geopotential height anomaly for the entire east Asia region (in July is 9 days,in June are 6 days,in August is 5 days). 3) Through analysis and comparison,it is found that a collection of 12 members (3 day moving) can get a stable circulation forecasting results in CFSv2. 4) The CFSv2 monthly circulation predictive skills in the entire East Asia region largely depend on the prediction level in the mid-high latitude area in the summer of 2015. The prediction skills in the mid-high latitude area (from 4 days in advance can reach a high level and maintain stable for June to August) are higher and more stable than in the tropics.