Abstract: To estimate the accuracy of AIRS retrieval products in the region of China, the surface air temperature derived from AIRS are compared with those obtained from 540 meteorological stations in the period of July 2004 and January 2005. The atmospheric temperature and humidity profiles derived from AIRS are also compared with T213 numerical forecasting products. Then the errors in AIRS atmospheric pro ducts are analyzed. Furthermore, the surface air temperature and atmospheric tem perature and humidity profiles derived from AIRS are assimilated into the MM5 nu merical model by analysis nudging, to investigate their effects on snowfall, and water vapor, vertical velocity and divergence fields as well as microphysical processes. The preliminary results are as following: (1) The errors in the surface air temperature retrieved from AIRS are systematic errors. The influence of terrain altitude can cause a negative deviation of 10 K. Also, the effects of land surface are significant, for the snow cover in winter can bring about a positive deviation of 5 K. (2) By comparing the atmospheric profiles derived from AIRS with those from T213 numerical products, it is found that the difference in temperature can reach 2 K, and the difference in relative humidity can reach 25%. (3) The numerical experiment results show that the use of AIRS atmospheric produ cts in the initial assimilation is able to improve the initial moisture field an d thus the snowfall prediction to some extent.