Theoretical models of depression have hypothesized phase or amplitude alterations in circadian measures including the sleep-wake rhythm. Whether such abnormalities occur only during the depressed state, or whether they persist into recovery, is less clear. We investigated the circadian pattern of unintended sleep episodes during 36 h of constant wakeful bedrest in two groups: 26 drug-free patients whose depression remitted following psychotherapy treatment, and a contrast group of 17 healthy young adults. The contrast group was not matched for age or gender. Both remitted depressed and contrast groups showed statistically significant linear and quadratic trends in the number of unintended sleep episodes, indicating monotonic and circadian influences across the study interval. We found no significant group differences in the pattern of sleep episodes. The number and timing of sleep episodes did not correlate significantly with core body temperature amplitude or timing or with baseline sleep duration or efficiency. The results do not support the hypothesis of a phase or amplitude change in sleep propensity as a trait abnormality in depression. However, the inclusion of a moderately depressed out-patient cohort, which showed only minor sleep changes and normal temperature profiles even while depressed, may have biased against finding significant differences.