For the purpose of pursuing precise source localization and reducing the impact of location accuracy brought by the initial value, an iterative algorithm which can converge robustly is proposed in this paper for the source localization using time-difference-of-arrival (TDOA). The new algorithm firstly introduces the maximum likelihood method to determine the objective function, and then uses the Newton method to find the source location. With the problem of the ill-condition Hessian matrix, the algorithm uses the Regularization theory to construct a symmetric definite Hessian matrix, which ensures the robust and the efficient of the algorithm. Experiments results show that this new algorithm is robust to the initial value, and is still able to ensure its convergence even with an inaccurate initial value of large error compared with the classical Newton method, and then compared with some other closed-form source location methods, the new algorithm has better accuracy in large noise levels.