This study used the soil moisture product of ESA CCI (the Climate Change Initiative)and combined land surface parameters, brightness temperature, soil texture, terrain, rainfall and other auxiliary data, to establish a downscaling model for the soil moisture. Based on the moving window method, this model combined linear and nonlinear techniques. It was applied to downscaling the CCI data of soil moisture over China from 2003 Jan. to 2013 Dec., in order to improve its spatial resolution. The downscaled data will be more suitable for the agriculture, hydrological simulation and other disciplines. Because ESA CCI product is not available over the whole China in winter, we further conducted a merge of ESA CCI product and the ERA-Interim product of soil surface moisture. Finally, we obtained the monthly soil moisture product over China with 1 km spatial resolution from 2003 to 2013. We further used in-situ observations to validate the product and found that this method improved not only the spatial resolution and the coverage, but also the precision.