Abstract: For poor and unintuitive results of traditional optical imaging in underwater concrete pile inspection in complex environments, a fast 3D reconstruction method for underwater defective piles based on multibeam image sonar is proposed. An underwater robot with two orthogonally oriented multibeam sonars collects sonar data at various positions and depths around the pile. 3D reconstruction is achieved via image preprocessing, defect contour feature extraction, multi-view contour stitching, and multi-depth contour stacking. Experimental results show low reconstruction errors, precise defect localization, and high reliability for depressions, which validates the method’s effectiveness in underwater bridge structure inspection.