The cerebral glucose utilization rate was studied for 27 normal volunteers with 18F-deoxyglucose positron emission tomography (PET). The scanner has a spatial resolution of 6-7 mm and contains corrections for scatter, attenuation, and random coincidences. The lumped constant (tracer-to-glucose dynamic uptake ratio) was determined by comparing the average global uptake of tracer in representative slices with average glucose utilization rates measured by the Kety-Schmidt method as reported in the literature. The resulting value of 0.50 is in excellent agreement with a recent direct determination done by arterial and jugular bulb blood sampling. Gray and white matter values of glucose utilization in various areas of the brain were determined by placing small regions of interest over various cortical, basal, and white matter structures. These values are within 20% of published autoradiographic data on the macaque monkey. The average ratio of gray to white glucose utilization was 2.9, compared with a range of 3-5 for the monkey study and 1.6-2.2 reported in previous PET studies. The effect of instrumental errors on the results is analyzed and discussed.