During prolonged starvation, brain energy requirements are covered in part by the metabolism of ketone bodies. It is unknown whether short-term starvation of a few days' duration may lead to reduced brain glucose metabolism due to the change toward ketone body consumption. In the present study we measured the cerebral metabolism of glucose and ketone bodies in nine healthy volunteers before and after 3.5 days of starvation. Regional glucose metabolism was measured by dynamic positron emission tomography using [18F]2-fluoro-2-deoxy-D-glucose. The mean value of K1* in gray and white matter increased by 12% (p < 0.05), whereas k2* and k3* were unchanged compared with control values. Regional glucose metabolism in cortical gray matter was reduced by 26% from 0.294 +/- 0.054 to 0.217 +/- 0.040 mumol g-1 min-1 (p < 0.001). White matter glucose metabolism decreased by 27% (p < 0.02). The decrease was uniform in gray and white matter with regional decreases ranging from 24 to 30%. A determination using Fick's principle confirmed the reduction in glucose metabolism yielding a decrease of 24% from 0.307 +/- 0.050 to 0.233 +/- 0.073 mumol g-1 min-1 (p < 0.05), whereas CBF did not change (0.57 +/- 0.07 vs. 0.57 +/- 0.06 ml g-1 min-1). The global net uptake of beta-hydroxybutyrate increased 13-fold from 0.012 +/- 0.024 to 0.155 +/- 0.140 mumol g-1 min-1 (p < 0.05). Net uptake of acetoacetate and net efflux of lactate and pyruvate did not change significantly during starvation. The present study shows that the human brain adapts to the changes in energy supply as early as 3 days following initiation of starvation, at which time ketone bodies account for approximately one-fourth of the cerebral energy requirements.