To determine whether insulin or noninsulin-dependent diabetes mellitus affects brain glucose metabolism, brain glucose utilization was studied in the basal state and during hyperinsulinemic euglycemic glucose clamps in nondiabetic and diabetic Pima Indians by positron emission tomography with 2-[18F]fluoro-2-deoxy-D-glucose (18FDG). Glucose utilization in 75 brain areas was determined by analysis of single scans and by least squares estimation of the rate parameters for the FDG model; these data were compared to results in normal caucasian volunteers. No effect of ethnicity or diabetic status on brain glucose utilization was observed. During the hyperinsulinemic clamps (mean insulin, 11,708 +/- 3,026 pmol/L), clearance of 18FDG from blood was accelerated, and accumulation of brain radioactivity was reduced. However, glucose utilization by the brain was identical to results during sham glucose clamps (mean insulin, 204 +/- 56 pmol/L) performed in the same patients. During the studies with hyperinsulinemia, k4 (representing loss of tissue radioactivity) was increased in most brain areas (mean increase, 0.0031 +/- 0.0018 min-1; P less than 0.02). The possible mechanisms for this effect are multiple, and the physiological significance, if any, is unknown. Further studies of the effects of insulin on brain glucose metabolism are needed.