In view of the problem of three-dimensional location of mobile terminals in indoor non-line-of-sight wireless environment, a three-dimensional location algorithm has thus been proposed based on linear regression and least square method in non-line-of-sight wireless environment. Firstly, based on the positively biased measurement error of arrival time (TOA) in wireless communication environment, the linear regression is used to estimate the linear relationship between the non-line-of-sight measurement distance error and the real distance between the test terminal and the base station. Then, combined with the linear relationship, the least square method is used to locate the 3D position of mobile terminal. The experimental results show that the proposed algorithm is characterized with a high positioning accuracy, with the maximum positioning error no more than 2 m. Moreover, in the non-line-of-sight environment, the proposed positioning algorithm does not need the prior knowledge of TOA,s time delay distribution, with the positioning accuracy better than other two-stage least squares positioning algorithm based on TOA.