The efficacy of [14C]glucose molecules labeled in various positions as tracers of regional cerebral glucose utilization (rCMRGlc) was examined in rats. Arteriovenous differences of different [14C]-glucose species and 14CO2 were measured across brain to determine the relative rates of 14CO2 loss. As anticipated, 14CO2 evolution decreased in the order: [U-14C]glucose greater than [2-14C]glucose greater than [1-14C]glucose greater than [6-14C]glucose. Release of 14CO2 from [6-14C]glucose was undetectable at 5 min and barely detectable at 10 min, and release from [1-14C]glucose, which includes the pentose phosphate pathway, was only slightly greater. rCMRGlc was measured with [1-14C]-,[2-14C]-, or [6-14C]glucose in 5-min experiments. The results of [1-14C]- and [6-14C]glucose were indistinguishable; no difference due to the activity of the pentose phosphate pathway was found. Both [1-14C]- and [6-14C]-glucose gave values similar to, but on the whole slightly higher than, [2-14C]glucose. It was concluded that when knowledge of total rCMRGlc is required, [6-14C]glucose is the labeled substrate of choice. When the experimental objective is measurement of energy metabolism, use of [1-14C]glucose avoids inclusion of the nonenergy-yielding pentose phosphate pathway.