The mechanism underlying improvement after total sleep deprivation (TSD) was studied in 14 major depressed patients. The suggestions that (1) circadian processes and/or (2) dimensions of arousal may play a role in the response to TSD were investigated. Diurnal variation of depressed mood and of mood- and arousal-related vocal parameters was studied in relation to the effect of TSD on depressed mood and vocal parameters. During 3 baseline days, during TSD and 2 days after TSD vocal parameters and depressed mood were assessed 6 and 3 times daily respectively. The mean fundamental frequency (frequency of vocal fold vibration, F0) (presumably reflecting aspects of arousal) as well as the range of the F0 (proposed to reflect sadness) showed a clear circadian pattern with a peak at about 4.00 p.m. TSD affected the circadian organization of the mean F0 and advanced the peak of the curve. After one night of subsequent sleep this effect disappeared. In addition, improvement after TSD coincided with an increase of the mean F0. The diurnal variation of mood before TSD predicted the mood response to TSD, whereas diurnal variation of vocal parameters did not. Moreover, circadian changes in vocal parameters were not related to changes in depressed mood. These findings suggest that the diurnal variations in mood and vocal parameters are regulated by different mechanisms. Data support the presumption that circadian as well as arousal processes are involved in the mood response to TSD. Circadian changes in vocal parameters due to TSD are not likely to reflect changes in the biological clock.