Previous studies have indicated that the European Center for Medium-Range Weather Forecasts (ECMWF) 40 years global reanalysis (ERA40) is better than the National Centers for Environmental Prediction-National Center for Atmospheric Research (NCEP-NCAR) Global Reanalysis (NCEP/NCAR) for describing the soil moisture over China, specially for the soil moisture variation due to snow melting and soil thawing at the end of spring and the beginning of summer. In this research, two soil moisture and temperature data sets at the end of May 1998 are compared and used as land initialization. Soil moisture and temperature in June 1998 is simulated using the mesoscale model MM5 to investigate the influence of soil condition on circulation and water cycle simulation. The results indicate that soil condition influences the circulation to some extent. Errors of zonal wind at 200 mb, geopotential height at 500 mb, northward meridian wind and air temperature at 850 mb simulated by using ERA40 soil data are less than that by using NCEP/NCAR soil data. Precipitation simulation by using ERA40 soil data is improved both in the higher spatial correlation coefficient and error, as compared with that by using NCEP/NCAR soil data. The evolution of zonal mean precipitation simulated by using ERA40 soil data is closer to the observation than that simulated by using NCEP/NCAR soil data.