The Weather Research and Forecasting model version 3. 6(WRFV3. 6) is used to investigate the impact of soil moisture on high-resolution simulation of the short-range high temperature weather occurred during 22-23 and 29-30 July 2003. Experimental results indicate that:(1) the simulated surface air temperature (SAT) SAT is highly sensitive to soil moisture perturbation. SAT increases (decreases) significantly in response to soil moisture decrease (increase). Meanwhile, the influence of soil moisture on SAT strongly depends on model resolution;(2) characteristic changes in SAT in response to soil moisture perturbations are similar in results of experiments at different model resolutions, and the model performance for SAT simulation improves with increased resolution;(3) changes in SAT are directly influenced by simulated sensible and latent heat fluxes in the sensitivity experiments;(4) soil moisture perturbation affects SAT simulation through altering the physical processes important for the development of high temperature weather. Among these processes, the convection (advection) exerts a warming (cooling) effect during the entire day, and the effect becomes weak with increases in soil moisture during the daytime. Under dry soil conditions, the diabatic heating plays a dominant role and its effect intensifies during the daytime;the effect of diabatic cooling weakens and weaker than the adiabatic heating of the convection process during the night-time. The above results show that soil moisture is important in the simulation and prediction of high temperature weather. Thereby it is feasible to improve the model ability of simulating and predicting high temperature using ensemble forecast with soil moisture perturbations.