Atmospheric Motion Vectors (AMVs) can supply plenty of useful information for numerical weather prediction. With the improvement in the image navigation, data calibration and derivation algorithm, the quality of FY2E is expected to be improved. Therefore, it is necessary to evaluate the improvement of FY2E AMVs for the analysis field and precipitation forecast in GRAPES (Global/Regional Assimilation and Prediction System) of China. In this study, the old and the reprocessed FY2E AMVs are used to analyze the characteristics of their horizontal and vertical structures and applied to GRAPES3Dvar Global Assimilation Prediction System to compare their differences on the assimilation and prediction. The experiments using the data collected in August 2013 show some encouraging results, which show neutral to positive impact on wind analysis field, especially in high levels. Furthermore, due to the improvement of the initial fields for the model prediction, the performance of the anomaly correlation coefficient (ACC) and root mean square error (RMSE) slightly improved. Especially the observation error of the reprocessed FY2E is lower than the old from 600 hPa to 200 hPa, which needs a further investigation. Conclusively, the reprocessed FY2E AMVs have more positive impact on wind assimilation and forecast improvement in GRAPES.