Abstract: Atmospheric Motional Vector (AMV) derived from continuous satellite images, their derivations and applications in Numerical Weather Prediction, weather analyses and forecasting as well as the recent research development are introduced and reviewed. At present, the algorithms of AMV derivation are basically identical among different satellite data processing centers. AMV image matching is used with correlation method, while AMV height assignment is physically based on one window channel and one absorption channel. The quality of AMVs derivation depends on the entire procedure of the data processing. Therefore, it is very important to guarantee the high performance of each step in data processing. Height assignment is still a challenging issue at present. Since physical height assignment does not turn out ideal results, attentions ought to be paid to the test on geometric way. Geostationary AMV gives positive contribution in tropics and southern hemisphere in NWP. Because of the lack of various types data in the polar zones, polar AMV has positive feed back in NWP. Feed back from MODIS polar winds is quite encouraging. Overlapping display of AMVs and satellite images give forecasters a unique tool at weather analysis and forecasting. It is especially useful for diagnosis of major rain belt, subtropical ridge, severe convections and tropical cyclones.