Abstract: The Community Land Model version 3.5(CLM3.5) is used to investigate the impact of underlying surface change on soil moisture (SM) simulation over Inner Mongolia.The model is driven by CLDAS (the atmospheric forcing data of CMA Land Data Assimilation System) outputs for the period of 2011 to 2013.Four 3-year (2011-2013) experiments with different surface information,i.e.the soil texture dataset from the Second National Soil Survey on China (SNSS) and CLCV (Chinese land cover derived from vegetation map)(CLM-new),the dataset from FAO and CLCV (CLM-clcv),the dataset from SNSS and the MODIS (CLM-snss),and the dataset from FAO (Food and Agriculture Organization of United Nations) and MODIS (CLM-ctl),are conducted.Results of SM simulations from the four experiments are compared to in-situ measurements of soil moisture collected at 46 stations.It is found that the observed temporal variations of SM can be better simulated in all the three experiments (CLM-new,CLM-snss,and CLM-clcv) compared to that in the control experiment (CLM-ctl).Major improvements are found over Hulunbuir in the CLM-clcv simulation,while improvements over large areas except Hulunbuir have been found in the CLM-snss simulation.Systematic overestimation of soil moisture at each soil layer is found in the CLM-ctl simulation.Among the four experiments,the CLM-new performs best and can well reproduce the spatial-temporal patterns of soil moisture in Inner Mongolia.The high correlation coefficient between the CLM-new simulation and observations and small bias and root mean square error indicate that this experiment can well simulate soil moisture in Inner Mongolia.