In order to effectively improve the accuracy of 3D ranging node localization in wireless sensor networks, a wireless sensor network node localization algorithm is proposed based on threedimensional ranging and LS-LM. Firstly, the tetrahedral volume formula of Helen’s formula is used, combined with the 2D transformation of the vector rotation on the plane, the estimated distance and direction can be obtained when only three anchor nodes are needed. Secondly, for the case where the number of anchor nodes is greater than 3, the LeastSquare LevenbergMarquardt (LS-LM) algorithm is used to optimize the estimation position of the target node. The experiments verified the feasibility of the proposed algorithm in threedimensional localization of wireless sensor networks. The simulation results show that the performance of the proposed algorithm is better and the localization accuracy could be improved by about 8% as compared with the traditional 3D LM and similar algorithms.