Psychometric functions were obtained for three listeners for the interaural conditions Nm-Sm and N0-Sa' using seven different noise spectrum levels. In addition, data were collected on the N0-S,r condition with the spectrum level in one ear smaller than that in the other ear. In these N0•-S,/conditions, the signal-tonoise ratio was kept equal in the two ears. The masking-level differences (MLD's) for the N0-S,r conditions declined gradually as the level of the noise was reduced, and for small spectrum levels, they were essentially the same as the MLD obtained when the masker was turned off. The MLD's declined more rapidly for the N0'-S,r • conditions, but a small MLD was obtained even when the spectrum level in one ear was 60 dB smaller than that in the other ear. For the Nm-Sm conditions, masking increased linearly with increases in spectrum level, but for N0-S,r, the relation was not linear--a further indication that the basis for detection is different in MLD and non-MLD conditions. The results are discussed in terms of "internal noise," the effect of which is to alter the effective interaural correlation as the masker level is reduced. Signal: 400 cps, 250 msec; masker: continuous, wide-band noise; method: two-interval forced choice. the signal is presumed to be greater, on the average, than it is on those trials or observation intervals containing noise-alone, and the listener presumably makes his decisions accordingly. For the MLD conditions, Jeffress followed a proposal of Webster (1951), which assumed that the basis for detection is the interaural time shifts between the signal plus noise in one ear and the signal plus noise in the other ear. No such time shifts are possible in the non-MLD conditions. The model proposed by Jeffress is able to account for many of the data obtained on MLD's. It is important, therefore, that the weaknesses of the model be closely examined, for it is possible that with minor modifications the model would be able to account for even more data. But, until detailed information is available on the inadequacies of the model in its present form, the nature of such modifications will not be obvious. The experiment reported here was designed to investigate one of the weaknesses of the Jeffress model.A small set of symbols is used to characterize the various interaural listening conditions. The masker, typically white noise, is designated N, and the signal, typically tonal, is designated S. If the interaural phase shift of the noise or of the signal is 0 ø, it is known as NO or SO, respectively. If the interaural phase shift is 180 ø, N•r or S•r is used. A monaural stimulus is characterized with an m. Thus, the monotic listening con-Redistribution subject to ASA license or copyright; see http://acousticalsociety.org/content/terms. Download to IP: 137.189.170.231 On: Sat