Abstract: The tests on sensitivity of long wave radiation (LWR) to at mospheric changes using RRTM radiative parameterization have been made. The result shows that the reduction effect of high cloud on the OLR, 300 and 500 hPa surface net long wave (LW) flux is larger than middle and low cloud. The reduction effect of low cloud on the 850 hPa and the surf ace net LW flux is larger than middle and high cloud. T he effect of clouds on LWR cooling rate is the largest in the presence of clouds, and rapidly decreases under the cloud levels, and is not influenced above the cloud levels. Downw ard LW fluxes at ground surface would decrease or increase, but net LW fluxes would reverse with the decrease or increase of water vapor content, also the closer to the ground surface, the larger the effect of water vapor changes on LWR. The effect of ozone on LWR is smaller relative to cloud and water vapor. This paper discusses two cases in different season of applying RRTM scheme to forecast LWR in WRF model and gives the results of forecasting and verifying all over China in the two days using the NCEP/AVN analysis data. The results show that OLR and 500 hPa surface net LW flux is very correlative to geopotential height field, but net LW flux at ground surface is also influenced by topography to a large extent. In addition, the errors of LWR increase with forecast time, but they possess the character of diurnal and seasonal variation, the forecast errors of net LW fluxes at ground surf ace are a bit smaller in the night time than in the day time, and the forecast error growth of LWR with forecast time is larger in summer and aut umnt han in winter.