Gravity matching navigation is one of key technologies for underwater navigation. Terrain contour matching (TERCOM), as a sequence matching method, is insensitive to the error of the initial position and has high robustness. However, it suffers from the matching low-efficiency and out-of-domain mismatches due to the grid traversal search and the domain boundary limitation, which affects TERCOM underwater navigation ability. To address these problems, a novel domain-center adaptive-transfer matching method (DAMM) is proposed. It first uses the optimal positioning of horizontal-and-vertical pre-matching lines to realize the adaptive transfer of the domain center from the INS-indicated ending point to the on-line optimal matching position. Then, a new re-matching grid domain based on this on-line transferred domain-center index value is adaptively generated, and the repositioning and update for the optimal matching position are performed by following the optimization principle of the matching indexes, which contributes to improving the matching efficiency and out-of-domain positioning reliability of underwater gravity matching navigation. Experimental results verify the outstanding underwater navigation performance of DAMM on the matching efficiency and out-of-domain positioning reliability. For three different gravity tracks, the DAMM matching efficiency, compared with TERCOM, is improved by 71%, while the number of out-of-domain mismatches and positioning accuracy are improved by more than 92% and 84%, respectively, on average.