Abstract: The feedback process between soil and atmosphere is very important in the climate system. Most researches on the interaction depend on the model outputs due to the scarce observation dataset of the soil moisture. Heavy flood event occurs in the mid-lower Yangtze basins in 1998. In winter and spring of that year, the soil moisture in Yangtze-Huaihe River basins is significantly wet. The concerned question is whether the abnormal soil moisture in prophase could provide valuable signal to summer rainfall. By using the improved regional climate model (RegCM_NCC), modeling study is undertaken to investigate the effects of the initial abnormal soil moisture in spring in Yangtze-Huaihe River basins (27°—35°N, 110°—120°E) on the regional climate. Based on the control run with climate values of soil moisture, two sensitivity experiments are carried out with "wet" and "dry" soil in the initial integration, respectively. Results indicate that varying the initial soil moisture in the key region shows significant influence on regional rainfall, wet soil results in the increase of local rainfall, humidity and evaporation near the ground, but the surface air temperature decreases rapidly, with the maximum value of 1.5 ℃. The experiment results of dry soil are converse to that of the wet soil. The regional climate response mainly depends on the changing surface radiation equivalent and land-atmosphere energy flux. The influences of initial abnormal soil moisture are remarkable in the first month, which can persist for several months, but the intensity deceases gradually. The change of air temperature is more remarkable and can persist longer time than the rainfall. In the mean time, the regional climate response to the abnormal soil moisture isn't confined in the key region, but also affects the rainfall and temperature in other regions due to the sub-circulation and large-scale advection. It can also cause the air temperature and humidity change at upper level through turbulent transfer. Thus, soil moisture has obvious effects on the regional-scale climate, but most of the present models can not describe the real-time variation of soil moisture due to the reason that only the climate virtual value of the soil moisture are used, which may be a factor of the modeling bias of the rainfall. The role of soil moisture would be attached much attention in the regional short-term climate research and prediction.