Non-line-of-sight(NLOS) propagation will decrease the localization accuracy if it is not taken into account in the mobile localization algorithm design, therefore, this paper presents an interacting multiple model-improved Kalman filter(IMM-IKF) based mobile node localization algorithm for the WSN. The position of unknown node is estimated under the interacting multiple model frame. The improved Kalman filter is employed to filter the estimated locations. The transition between the LOS and NLOS is described by two state Markov models. Two different filtering results are combined according to the likelihood probability to provide the final location estimation. Simulation results demonstrate that the IMM-IKF algorithm can overcome the NLOS errors and achieve high location accuracy.