Abstract: The control experiment and two assimilation experiments of WRF regional mesoscale model prediction are verified for typhoon Haitang (0505). Based on the WRF-3DVAR system, the FY-2C infrared and water vapor cloud motion winds are assimilated into WRF model. The assimilation data of one assimilation test are the cloud motion winds before quality control and those of the other are cloud motion winds after quality control. The effect of the cloud motion winds on the prediction of rain and wind field is analyzed through comparing the three tests. Results show that the assimilation of the quality-controled cloud motion winds into the initial fields has positive impact on prediction of the area and intensity of rain, and also indicate that different grades of Ts score are improved by far than the other tests. The assimilation makes improvement in the forecast of wind. The same approach as that for Haitang is taken to simulate other two typhoon cases and the Ts scores of the forecast rainfall are given. And the results for the two typhoon cases show similar conclusions. Therefore, assimilating cloud motion winds with reasonable control and selection can well improve the descriptive possibility of wind field in WRF model, supply the mesoscale information in the initialization field to improve the simulation result of precipitation and wind field and have positive impact on the forecasting ability of WRF model finally.