To discuss the application effectiveness of grey system theory in sound quality prediction modeling, 30 radiation noise samples of 5 different 3 tonnage forklifts under the conditions of idle and rated speed were taken as the research object, and subjective evaluation experiment was carried out with grading method taking annoyance as subjective evaluation index; The primary objective Psychology acoustical parameters were calculated with ArtmiS, and the loudness, sharpness, roughness and fluctuation were selected as the main analysis object; By using the grey system theory, the correlation between the annoyance and the objective parameters of psychology was analyzed, and the correlation coefficient was obtained; It was proved that the correlation between the objective parameters and the annoyance was high. Based on the GM(0,N) model of grey system theory, the prediction model of annoyance was established, and the error test of the prediction model was carried out; Results showed that the prediction model based on grey system theory had higher accuracy, that was to say and the prediction of value was more close to human"s subjective feelings.