A soil moisture assimilation scheme is developed, which is based on the ensemble Kalman filter (EnKF), an unsaturated soil water flow model and the Variable Infiltration Capacity (VIC) land surface hydrologic model. To validate and verify the scheme, several assimilation experiments with in-situ observations and infiltration derived from the VIC model driven by observed meteorologic forcings are presented. The results demonstrate that the assimilation scheme can retrieve the true soil moisture profile, improve the estimation of soil moisture, and capture the daily, dekad, monthly and seasonal variation of soil moisture, which shows that the scheme is reasonable. Comparisons of the assimilation experiments with the algorithms EnKF and Extended Kalman Filter (EKF) show that the EnKF scheme with the suitable ensemble size is easy to implement and is a little superior to the EKF scheme for soil moisture assimilation, but the computational cost for the EnKF scheme is higher than that for the EKF scheme.