Nocturnal sleep restriction and compensation with daytime naps is common in today's society. In a between-participants design, we examined the effects of chronic (10 nights) sleep restriction on 24 h plasma melatonin profiles in humans. Following a baseline period with 8.2 h time in bed (TIB) for sleep, participants were randomized to a control (8.2 h TIB) or sleep-restriction condition (4.2 h TIB), with and without diurnal naps. Sleep restriction was achieved via delaying bedtime and advancing wake time by 2 h each relative to the baseline sleep period. Participants were maintained in a controlled, time isolated laboratory environment throughout the protocol, with light levels below 40 lx at all times. Twenty-four hour plasma melatonin profiles were assessed at baseline and at the end of the sleep-restriction period, with subjects maintained in a constant posture protocol. Compared with the baseline assessment and the 8.2 h TIB control group, a significant phase delay in melatonin onset (1.2 ± 0.9 h) occurred in all sleep-restriction (4.2 h TIB) groups (P < 0.05). There was no evidence of a phase advance or shortening of the period of melatonin secretion associated with the advanced waking time. These results suggest that nocturnal light and dark exposure may be more potent in effecting circadian phase shifts than exposure to morning light, at least in conditions of controlled, dim lighting in the laboratory.