W e have studied the effect of prolonged hyperinsulinemia and hyperglycemia on serum leptin levels in young nonobese males during 72-h euglycemic-hyperinsulinemic and hyperglycemic ( approximately 8.5 and 12.6 mM) clamps. Hyperinsulinemia increased serum leptin concentrations (by RIA) dose-dependently. An increase in serum insulin concentration of > 200 pM for > 24 h was needed to significantly increase serum leptin. An increase of approximately 800 pM increased serum leptin by approximately 70% over 72 h. Changes in plasma glucose concentrations (from approximately 5.0 to approximately 12.6 mM) or changes in plasma FFA concentrations (from < 100 to > 1,000 microM) had no effect on serum leptin. Serum leptin concentrations changed with circadian rhythmicity. The cycle length was approximately 24 h, and the cycle amplitude (peak to trough) was approximately 50%. The circadian leptin cycles and the circadian cycles of total body insulin sensitivity (i.e., GIR, the glucose infusion rates needed to maintain euglycemia during hyperinsulinemic clamping) changed in a mirror image fashion. Moreover, GIR decreased between Days 2 and 3 (from 11.4+/-0.2 to 9. 8+/-0.2 mg/kg min, P< 0.05) when mean 24-h leptin levels reached a peak. In summary, we found (a) that 72 h of hyperinsulinemia increased serum leptin levels dose-dependently; (b) that hyperglycemia or high plasma FFA levels did not affect leptin release; (c) that leptin was released with circadian rhythmicity, and (d) that 24-h leptin cycles correlated inversely with 24-h cycles of insulin sensitivity. We speculate that the close positive correlation between body fat and leptin is mediated, at least in part, by insulin.