T o define the roles of circadian rhythmicity (intrinsic effects of time of day independent of the sleep or wake condition) and sleep (intrinsic effects of the sleep condition, irrespective of the time of day) on the 24-h variation in glucose tolerance, eight normal men were studied during constant glucose infusion for a total of 53 h. The period of study included 8 h of nocturnal sleep, 28 h of continuous wakefulness, and 8 h of daytime sleep. Blood samples for the measurement of glucose, insulin, C-peptide, cortisol, and growth hormone were collected at 20-min intervals throughout the entire study. Insulin secretion rates were derived from C-peptide levels by deconvolution. Sleep was polygraphically monitored. During nocturnal sleep, levels of glucose and insulin secretion increased by 31 +/- 5% and 60 +/- 11%, respectively, and returned to baseline in the morning. During sleep deprivation, glucose levels and insulin secretion rose again to reach a maximum at a time corresponding to the beginning of the habitual sleep period. The magnitude of the rise above morning levels averaged 17 +/- 5% for glucose and 49 +/- 8% for calculated insulin secretion. Serum insulin levels did not parallel the circadian variation in insulin secretion, indicating the existence of an approximate 40% increase in insulin clearance during the night. Daytime sleep was associated with a 16 +/- 3% rise in glucose levels, a 55 +/- 7% rise in insulin secretion, and a 39 +/- 5% rise in serum insulin. The diurnal variation in insulin secretion was inversely related to the cortisol rhythm, with a significant correlation of the magnitudes of their morning to evening excursions. Sleep-associated rises in glucose correlated with the amount of concomitant growth hormone secreted. These studies demonstrate previously underappreciated effects of circadian rhythmicity and sleep on glucose levels, insulin secretion, and insulin clearance, and suggest that these effects could be partially mediated by cortisol and growth hormone.