Analyses of the control of glucose metabolism by insulin have been hampered by changes in bloog glucose concentration induced by insulin administration with resultant activation of hypoglycemic counterregulatory mechanisms. To eliminate such mechanisms, we have employed the glucose clamp technique which allows maintenance of fasting blood glucose concentration during and after the administration of insulin. Analyses of six studies performed in young healthy men in the postabsorptive state utilizing the concurrent administration of [14C]glucose and 1 mU/kg per min (40 mU/m2 per min) porcine insulin led to the development of kinetic models for insulin and for glucose. These models account quantitatively for the control of insulin on glucose utilization and on endogenous glucose production during nonsteady states. The glucose model, a parallel three-compartment model, has a central compartment (mass = 68 +/- 7 mg/kg; space of distribution = blood water volume) in rapid equilibrium with a smaller compartment (50 +/- 17 mg/kg) and in slow equilibrium with a larger compartment (96 +/-21 mg/kg). The total plasma equivalent space for the glucose system averaged 15.8 liters or 20.3% body weight. Two modes of glucose loss are introduced in the model. One is a zero-order loss (insulin and glucose independent) from blood to the central nervous system; its magnitude was estimated from published data. The other is an insulin-dependent loss, occurring from the rapidly equilibrating compartment and, in the basal period, is smaller than the insulin-independent loss. Endogenous glucose production averaged 1.74 mg/kg per min in the basal state and enters the central compartment directly. During the glucose clamp experiments plasma insulin levels reached a plateau of 95 +/-8 microU/ml. Over the entire range of insulin levels studied, glucose losses were best correlated with levels of insulin in a slowly equilibrating insulin compartment of a three-compartment insulin model. A proportional control by this compartment on glucose utilization was adequate to satisfy the observed data. Insulin also rapidly decreased the endogenous glucose production to 33% of its basal level (0.58 mg/kg per min), this suppression being maintained for at least 40 min after exogenous insulin infusion was terminated and after plasma insulin concentrations had returned to basal levels. The change in glucose utilization per unit change in insulin in the slowly equilibrating insulin compartment is proposed as a new measure for insulin sensitivity. This defines insulin effects more precisely than previously used measures, such as plasma glucose/plasma insulin concentration ratios. Glucose clamp studies and the modeling of the coupled kinetics of glucose and insulin offers a new and potentially valuable tool to the study of altered states of carbohydrate metabolism.