Abstract:The observed yields of winter wheat from 123 agrometeorological observation stations and the announced yield of winter wheat at county level， where the observation station is located， are used to integrate the observed yield and announced yield at provincial and national levels by the proportion of winter wheat planting area. And the observed yield and announced yield sequences of winter wheat are compared at provincial and national levels. The observed and announced yields of winter wheat at national level are predicted based on the climate suitability index forecast method. Also, the forecast accuracy of different yield sequences is analyzed. The results show that the observed yields are higher than the announced yields at provincial level in all provinces. The correlation coefficient between the observed and announced yields is good in each province and has passed the significant test except in Xinjiang. The correlation coefficient between the observed yield and announced yield at national level reaches 0.97, and the observed yield could reflect the characteristic of announced yield. Besides, the percent of consistency statistics of trend meteorological yield of observed and announced yields remains good at national level, so it is suitable to carry out yield prediction, but it is unsuitable to carry out yield prediction at province level due to the low percent. The accuracy of different yield sequences in forecasting their own sequences is high and the accuracy of announced yield is higher than the observed yield. However, the accuracy of the forecast conversion of the announced yield by using the observed yield would be reduced. Conclusively, it is feasible to carry out yield forecast at national level based on the observed yield series because of the real time, objectivity and representative of the observed yield. At the same time, the new yield series could provide new data support for yield prediction.