Studies comparing glucose tolerance in the morning vs. that in the evening have suggested that time of day may influence glucose regulation. To examine the variation in glucose tolerance throughout the 24-h span, normal subjects were given an iv glucose infusion at a constant rate of either 5 or 8 g/kg.24 h during 30 h, and plasma levels of insulin and glucose were measured at 15-min intervals for the last 24 h of the infusion. The timing of initiation of the infusion was varied to differentiate effects of time of day from effects of duration of the infusion. A nocturnal elevation of glucose levels, culminating around midsleep and corresponding to an increase of about 15% above daytime levels, was observed in all subjects. The timing of this nocturnal maximum was not dependent on the rate of the infusion or on the time elapsed since the beginning of the infusion. Insulin levels did not show a consistent diurnal pattern. Both insulin and glucose exhibited large ultradian oscillations recurring at 100- to 150-min intervals. The amplitude of these oscillations increased with the rate of glucose infusion. These ultradian oscillations of glucose and insulin levels were temporally correlated, with a tendency for glucose pulses to lead insulin pulses by 15-30 min. These results demonstrate in normal subjects the existence of a diurnal variation in glucose tolerance distinct from the dawn phenomenon observed in diabetic subjects and indicate that spontaneous 100- to 150-min oscillations in peripheral glucose and insulin levels characterize stimulated pancreatic function, with the amplitude of the oscillations being dependent on the size of the stimulus.