Abstract: A weighted signal tensor subspace fitting algorithm is presented for vector hydrophone array based on higher order singular value decomposition (HOSVD). In this paper, the 3rd order tensor of the received signals from vector hydrophones array is modeled at first, then the signal tensor subspace is derived from HOSVD, and lastly, the DOA is estimated with the weighted signal subspace fitting. The 3rd tensor-based signal subspace estimation via HOSVD is a better estimate of the desired signal subspace than the subspace estimate obtained by the SVD of a matrix which exploits the structure inherent in the multi dimensional measurement data. Theoretical and simulation results show that the proposed method exhibits high resolution and robustness performance under scenarios of low signal noise ratio (SNR), non-correlative and tight-correlative signals with the same power.