The classical direction of arrival (DOA) estimation algorithms for coherent signals are mainly based on the spatial smoothing method and the subspace fitting algorithm. These methods suffer from small array aperture or heavy computation load. It is important to find a method for coherent signals with a large array aperture and small computation load. In this paper, a new method is proposed based on the vector reconstruction technique for coherent signals using the total least square-estimation of signal parameters via rotational invariance technique (TLS-ESPRIT) algorithm. The proposed method exploits all eigenvectors corresponding to the signal subspace and the forward-backward vector reconstruction technique to construct the receiving data matrix. The method is effective when both independent and coherent signals are incident on the array, and useful in practical signal environment. Theoretical analysis and simulation results show that the proposed method has low computation load, large array aperture and satisfactory estimation performance.