It is quite significant to investigate the law of perception concerning level-fluctuating noises when we consider that the levels of sounds which surround us in our daily life, such as music, speech, etc., are not fixed, but fluctuating at any moment. Accordingly, in this experiment, level-fluctuating noises were generated by systematically changing the physical intensity of noises by means of tapecontrolled system and their loudness was investigated by matching it with that of level-fixed noises. Both noises, level-fluctuating and level-fixed, were composed of 50 constituent white noises, on- and off-time of which were 72 msec and 34 msec respectively. One circuit of stimulus was composed of 5sec presentation of level-fluctuating noise, 1sec silent interval, 5sec presentation of level-fixed noise and 1sec silent interval. The levels of constituent noises of level-fluctuating noises were 66, 68, 70, 72 & 74dB in series 1: 62, 66, 70, 74 & 78dB in series 2: 58, 64, 70, 76 & 82dB in series 3: 54, 62, 70, 78 & 86dB in series 4: and 50, 60, 70, 80 & 90dB in series 5. 9 kinds of stimulus distribution as shown in Table 2b were applied to each series. By controlling the remote-control attenuator Ss matched the loudness of level-fluctuating noises with that of level-fixed noises, listening to both noises as many times as they liked. It may appear difficult to match level-fluctuating noises with level-fixed noises, but actually Ss reported that it was rather easy and intra-individual variance was small. As a result, it was found that there was the relation designated by equation (4)-(8) between the physical intensity and the loudness of level-fluctuating noises and that the loudness was expressed as a function of the arithmetic average of SPL in dB. The forms of equation (2)-(6) were found to be similar to the form of equation for predicting AL in Helson's AL theory, so a discussion was made about their relation. Further, Ss did not respond to the noises one by one which constituted level-fluctuating noises, but seized the constituent noises as a whole. Taking this fact into consideration, it was pointed out that it would be necessary to consider this problem with reference to “temporal quantum” of Bekesy, “chunk” of Miller, G. A., short-term memory and perceptual present. Besides, as it was found that the loudness of level-fluctuating noises systematically changed in accordance with the width of the level-fluctuating range, the phenomena of temporal masking might be thought to intervene in this problem. But when the duration of level-fluctuating noises are rather long as in this experiment, it is not quite adequate to explain the loudness of level-fluctuating noises by means of the phenomena of temporal masking alone. Any way, this experiment is still a preliminary one and the data are not enough to suggest a definite model. More detailed and expanded experiments would be necessary in future in order to investigate the factors which regulate the perception of level-fluctuating noises.