We have developed a method for the noninvasive estimation of regional tissue glucose utilization in humans that employs positron emission tomography (PET) and 2-(18F)fluoro-2-deoxy-d-glucose (FDG). Unlike other methods, the input function used in this method is obtained from the corrected time-activity curve of the descending aorta, not the left ventricle, because the descending aorta is relatively free of spillover from other organs and extends from the upper thorax to the lower abdomen. With this method the time-activity curve of the descending aorta must be corrected for the partial volume effect and the difference in counts between plasma and whole blood. Using the noninvasively obtained input function, regional tissue glucose utilization was calculated by Patlak graphic analysis. k1k3/(k2 + k3) was in good agreement with k1k3/(k2 + k3) calculated from the plasma input function by arterial sampling (r = 0.9995). These results suggest that the input function and regional tissue glucose utilization (not only of myocardium but also of other thoracic and abdominal organs) can be determined noninvasively.