To investigate the difference of tinnitus masking level and characteristics of masking effect between pure tone audiometry and Bekesy audiometry and to evaluate which is more effective and simple between these two tests, author did tinnitus masking to 55 persons(77 ears.) The results were summarized as follows : 1) 45 cases(81.8%) had tinnitus-inducing presumptive causes such as drug, noise, trauma of head and neck, inflammatory middle ear disease and so on. 56 ears(72.7%) accompanied hearing disturbance, among which 46 ears (59.8%) were sensorineural type and high frequency hearing loss was prevalent. 2) Pure tone threshold was 7.5-12.7dB higher than Bekesy one in each frequency (correlation coefficient, 0.926), and also pure tone masking level to tinnitus was 5.6-14.8dB higher than Bekesy one in each frequency (correlation coefficient, 0.867). The formula of regression line between sensation level of pure tone masking (PMSL, y) and Bekesy masking (BMSL, x) was y=1.115x+0.813 (correlation coefficient, 0.774). 3) The correlation coefficient between amplitude of Bekesy audiogram and Bekesy masking curve was about 0.605. 4) With pitch match · loudness balance test, the tinnitus frequencies were mostly 1kHz (53.5%) and 4kHz (29.6%) and the tinnitus loudness was 15-30dB (66.2%). The formula of regression line between Bekesy masking level(y) and loudness balance level(x) was y=0.339x+17.199 (correlation coefficient, 0.625). The results which the masking level with Bekesy audiometry is lower than that of pure tone audiometry (correlation coefficient, 0.867) suggest that the tinnitus masking with Bekesy audiometry could be applied for evaluation of tinnitus because it is more simple and effective than masking with pure tone audiometry.