In two recent articles, Hood and Wade (1982) and Weiler et al. (1981) have argued that the loudness of a steady tone does not appear to decline over time unless listeners are given a reference sound by which to judge loudness. The present experiments show, by the method of successive magnitude estimation, that listeners do not need a reference sound in order (1) to track accurately the decline in the loudness of a tone slowly decreasing from 60 to 40 dB SPL or (2) to track the loudness decline of a constant-intensity tone under special conditions that lead to adaptation. Since a reference sound is not needed by listeners to track a decline in loudness, and since Hood and Wade and Weiler et al. have found a decline with a reference but not without, it follows that adding a reference sound to a sustained sound must induce adaptation that otherwise does not occur. In support of that interpretation, the present measurements show that when subjects are told to ignore the same reference sound as used by Hood and Wade and by Weiler et al.--an increment every 30 s to a steady tone--loudness still declines. The bigger the increment (20 dB v. 5 dB) and the longer (5 s v. 1 s), the more loudness declines. Thus, loudness adaptation may be induced not only by a contralateral intermittent sound (Botte et al., 1982) but also by an ipsilateral intermittent sound. However, under normal listening conditions at levels above about 30 dB SL, loudness does not adapt.