Because soil thermal properties depend on soil moisture, soil temperature is strongly controlled by soil moisture. This provides a possibility to derive soil moisture from soil temperature measurements. Following this motive, this study develops a system to inversely estimate soil moisture profile from soil temperature profile. The forward model is a single-source land surface model to simulate land surface and subsurface processes. The cost function is a non-dimensional function to describe the discrepancy between measured and model-predicted values of soil temperatures. Besides the soil moisture profile, the estimated parameters also include soil hydraulic and thermal parameters. Applications at one synthetic case and three field cases (two in Tibet, and one in Iowa) show the inverse method can reproduce observed soil moisture when the soil properties can be described by soil hydraulic and thermal function used in this study. In addition, this method can simultaneously produce reasonable surface energy partition.