On-line quality control in automated welding operations is an important factor contributing to higher productivity, lower costs and greater reliability of the welded components. The sound signal of plasma are welding was acquired at high speed and investigated with the aid of computers, it is shown that the amplitude of the sound signal greatly varies with the variation of the status of the weld pool. The analysis of frequency domain indicated that the area below the curve of the low frequency band of the power density spectra, which may reflect the oscillation properties of the weld pool, is largest on the transition phase. The high frequency dominant peaks of the sound signal, which may relate to the plasma jet pulsation, is highest on the keyhole-formed phase. The method for detecting the behaviors of the weld pool keyhole was developed. The experimental results indicate that the acoustic emission is a usable, practical information source in the penetration quality detection of plasma are welding.