BACKGROUND Myocardial glucose uptake can be measured with [18F]fluoro-2-deoxyglucose (FDG) and positron emission tomography (PET). However, changes of myocardial metabolism may alter the ratio between the net rates of FDG and glucose uptake, known as the lumped constant. We tested the hypothesis that the variability of the lumped constant determined in animals explains the disagreement between human net myocardial glucose uptake calculated from aortocoronary sinus deficits and measured with PET. RESULTS In the three-compartment model of glucose transfer into cells, the lumped constant is a function of the relationship between the net and the unidirectional rates of uptake of glucose and glucose tracers such as FDG. Using this principle, validated in the human brain and the animal heart under experimental conditions, we estimated the lumped constant of the human heart by PET in 10 healthy men under several metabolic conditions established by altering the circulating insulin level during a euglycemic clamp and with somatostatin and heparin infusions. The lumped constant varied systematically between 0.44 and 1.35. At insulin levels below 100 pmol/L, free fatty acids were inversely related to serum insulin levels and the lumped constant increased linearly with serum insulin concentration. At insulin levels above 100 pmol/L, free fatty acids were suppressed and the lumped constant varied in inverse proportion to the insulin level. When the lumped constant was estimated in this manner, net myocardial glucose uptake agreed with that determined in previous measurements of blood flow and aortocoronary sinus deficit. CONCLUSIONS In the intact human organism, the cardiac lumped constant varies with the metabolic condition, as predicted from studies of the brain and animal heart under experimental conditions.