The objective of this project is to develop a knowledge-based ultrasonic inspection system (KBIS) for inspecting steel welds in submarine hulls. A key element of the KBIS system is a neural network classifier that is trained to classify a discontinuity as slag, porosity, lack of fusion or a crack. The software has been integrated into the commercial ABB Amdata ultrasonic inspection system and has been tested and validated for use by the U.S. Navy at locations including the Puget Sound Naval Shipyard, the Electric Boat Corporation and General Dynamics.