Although the human auditory perception is very complicated and non-linear device and it cannot be represented by the physical quantities only, the conventional noise source identification has been carried out based on the data represented in the sound intensity and/or sound pressure level. Because the conventional identification techniques are not counting for the psychoacoustic nature of the human hearing, the noise control engineers have often met the paradoxical results after taking the elaborate noise control measures: no change in the annoyance of noises although the overall A-weighted sound pressure level is reduced as desired initially. The purpose of this study is to bring the psychoacoustic concept into the practical noise control actions. In this study, expressions of sound quality metrics have been studied and the computerized evaluation routines were constructed for the noise source identification considering the psychoacoustic perception characteristics. First, the noise source identification in the viewpoint of sound power level and the A-weighted sound pressure level was performed and their limitations were investigated: sound intensity mapping and spectral analysis were utilized. Second, the procedures for the noise source identification in the viewpoint of the four major sound quality metrics were proposed: loudness, sharpness, roughness and fluctuation strength. Finally, the ordinary and the partial coherence function were adapted for evaluating the causality relationship of the vibro-acoustic transfer functions between the sound pressure at the receiver point and the sound pressure near the source or the surface vibrational velocity on the source. The applicability of the suggested idea was checked through the simple example of the baffled 4-loudspeaker system in which the a priori known signals were fed. Then, the present method was applied to the noise source identification of the commercial air compressor and the results showed that the method is very promising in practice. The present technique is thought to be useful in the control of machinery noise which contains many different sound quality features.