lnsulin action in vivo requires a cascade of individual steps including insulin transport across capillaries. Recepter binding and phosphorylation, translocation of glucose transporters to the membrane and intracellular metabolic events. The minimal model of insulin action assumes that glucose utilization is proportional to a "remote" insuln compartment (X (t)). If X (t) is closeIy related to interstitial insuIin, then transcapillary insulin transport is rate limiting for insulin action. We examined the relationsbip between plasma and interstitial (lymph) insulin, and X (t) predicted by the model. Right atrial blood and thoracic duct lymph were collected simultaneoualy during IVGTT (0,3 g/ kg glucose; 3 hrs; 29 samples) in conscious dogs (n=5). Plasma insulin peaked 5+-2 min (106-23 Pu/ml) after glucose injection; Iymph insulin rose slowly (22+-5 min) to lower peak (58+-9 Pu/ml), suggesting a barrier to the transcapillary msulin transport The dynamics of X (t) were remarkably similar to lymph insulin (r=0.95+- 0 .02, p<0.01), while X (t) was only weakly correlated with plasma insulin (r=0.65+-0.04, p<0.01). The intimate relationahip between the rate of glueose utilization and lymph insulin during the IVGT validates the minimal model, and suppolts the assumption that insulin transport tn the interstitium bathing insulin sensitive cells is the rate limiting step for the insulin action.