Abstract: On the base of the idea of matched-mode beamformer, a new matched-mode beamformer using total least square algorithm was proposed in this paper. Firstly, the normal mode depth functions are extracted by mode decomposition from the received dada in the vertical linear array (VLA). Then, the estimated normal modes are modified by total least square algorithm, resulting in estimating the normal mode coefficients more accurately. Finally, the matched-mode beamforming for localization is achieved according to the estimated mode coefficients. Numerical simulations in typical shallow water environments demonstrate the improvement of the proposed method. The simulation results indicate that the proposed method estimates the normal mode coefficients more accurately, and has characteristics of low sidelobe, high resolution, and good localization in compared with the matched-mode beamformer using least square algorithm in low-frequency far field environment.