Regional cerebral [11C]3-O-methyl-D-glucose ([11C]MeG) uptake kinetics have been measured in five insulin-dependent diabetic patients and four normal controls using positron emission tomography (PET). Concomitant measurement of regional cerebral blood volume and CBF enabled corrections for the presence of intravascular [11C]MeG signal in cerebral regions of interest to be carried out, and regional cerebral [11C]MeG unidirectional extraction fractions to be computed. Four of the five diabetic subjects were studied with their fasting plasma glucose level clamped at a normoglycaemic level (4 mM), and four were studied at hyperglycaemic plasma glucose levels (mean 13 mM). The four diabetic subjects whose fasting plasma glucose levels were clamped at a normoglycaemic level of 4 mM had mean fasting whole-brain, cortical, and white matter [11C]MeG extraction fractions of 15, 15, and 16%, respectively, values similar to those found for the four normal controls (whole brain, 14%; cortex, 13%; white matter, 17%). Mean regional cerebral [11C]MeG extraction fractions were significantly reduced in diabetic subjects during hyperglycaemia whether their plasma insulin levels were undetectable or whether they were raised by continuous intravenous insulin infusion. Such a reduction in [11C]MeG extraction under hyperglycaemic conditions can be explained entirely in terms of increased competition between [11C]MeG and D-glucose for the passive facilitated transport carrier system for hexoses across the blood-brain barrier (BBB). It is concluded that the number and affinity of D-glucose carriers present in the BBB are within normal limits in treated insulin-dependent diabetic subjects. In addition, insulin appears to have no effect on the transport of D-glucose across the BBB.