Most studies concerning with hearing ability have been so far done by using synthetic pure and complex tones. Moreover, these studies have been performed for judging pitch, loudness, tone color, tone length, etc., they can therefore be seemed to be a factor analytical research, or a microscopical method. The purpose of this work is to study on the comparison of hearing ability and to obtain acoustical information on educational media by using natural sounds having various tone characters, or by a macroscopical method. 48 paired sounds were prepared and used as problems. 36 problems were of "different" pairs and 12 were of "the same" pairs. After the presentation of a paired sound in a stereophonic mode, the subjects judged the difference and then pressed the number on their analyzers which is appropriated for their answers as follows : The same (1); Seemed to be the same (2) I Indistinguishable (3); Seemed to be different (4)', and, Different (5). In this work two experiments were performed. In Experiment I no sample sound was given. 100 normal-hearing subjects aged 15 to 48 were tested. In Experiment II sample sounds were given before the test in order to examine the effect of preliminary exercise on hearing ability. 78 normal-hearing subjects aged 15 to 56 were tested. Experiment II was performed by using the same problems as in Experiment I. The proportion of correct answers (PCA) was improved from 0.50 in Experiment I to 0.52 in Experiment II, and that of incorrect answers (PIA) was also improved from 0.28 to 0.23, respectively. The problems were made so as to preset various degrees of difficulty in judging the differences, and "the same" problems were introduced to check the authenticity of the subjects. Also, one problem was used twice in each experiment in order to check the reliability of the subjects and the problems. For the latter, the proportion of subjects who answered consistently to both problems was 62.5% in Experiment I and 72.7% in Experiment II. On the other hand, the proportion of subjects w ho answered inconsistently was 22.9% and 18.2%, respectively. The hearing ability of the latter can be said to be questionable in these experiments. The plots of "the same" problems dispersed homogeneously in the PCA-PIA pattern in Experiment I, while most of "the same" problems tended to be answerd incorrectly in Experiment II. From the data for the "different" problems, much information on tendencies in the discriminnation of sounds was obtained. Some of the results are described as follows : To judge the differences between homophasic and antiphasic sounds became more difficult as the phase conditions involved become more complex ; To detect the differences between normal pitch and 5% lower pitch in frequency was easier than that of a higher pitch of the same quantity ; On equalized tones for both "waterfall" and "violin", peaking of ±4dB at 2.8KHz and peaking of - 8dB at 700Hz or 11KHz could hardly be judged, but it could be deduced that it was easier to judge "violin" than "waterfall", with similarly equalized tones. In Experiment II, it could be seen that the mental set which selects information in judging the differences between two sounds might be obtained by the sample sounds, and that the improvement of hearing ability may be possible by means of training. Consequently, when human hearing is examined, whether sample sounds are given or not is a very important factor in measuring the thresholds of discrimination. This work clarified that hearing ability for the adults was superior to the students aged 15 to 17. This fact was explained by perception development and richness of sound experiences both in quality and quantity. The correlation between the results of the experiments and experience and/or environment with regard to sounds was also studied. For adults, some correlation was found.