Multichannel acoustic emission monitoring was examined for its ability to differentiate between sound and defective welds produced by submerged arc welding of steel plate. Postwelding acoustic emission signals were recorded and analyzed with respect to their maximum amplitude, the pulse duration and the frequency spectrum. Waveform analysis of acoustic signals in the postweld period, namely, maximum amplitude distribution, pulse width, and pulse frequency spectrum, has shown that there is a marked difference between signals resulting from sound and defective welds. This difference can be used to characterize the quality of a submerged arc weld.