In the process of indoor localiza-tion, the existence of the non-line of sight (NLOS) error will greatly reduce the local-ization accuracy. To reduce the impact of this error, a 3 dimensional (3D) indoor localization algorithm named LMR (LLS-Minimum-Re-sidual) is proposed in this paper. We first esti-mate the NLOS error and use it to correct the measurement distances, and then calculate the target location with linear least squares (LLS) solution. The final nodes location can be ob-tained accurately by NLOS error mitigation. Our algorithm can work efficiently in both in-door 2D and 3D environments. The simulation results show that the proposed algorithm has better performance than traditional algorithms and it can significantly improve the localiza-tion accuracy.