By means of emission computed tomography (ECT), we used 18F-fluorodeoxyglucose (18FDG) and 13N-ammonia (13NH3) as indicators of abnormalities in local cerebral glucose utilization (LCMRglc) and relative perfusion, respectively. The ECAT positron tomograph was used to scan normal control subject and 10 stroke patients at various times during recovery. In normal subjects, mean CMRglc was 5.28 +/- 0.76 mg per 100 gm tissue per minute (mean +/- SD; N = 8). In patients with stroke, mean CMRglc in the contralateral hemisphere was moderately decreased during the first week, profoundly depressed in irreversible coma, and normal after clinical recovery. Quantification was restricted by incomplete understanding of tracer behavior in diseased brain, but relative local distributions of 18FDG and 13NH3 trapping qualitatively reflected the increases and decreases as well as coupling and uncoupling expected for local alterations in glucose utilization and perfusion in stroke. Early after cerebrovascular occlusion there was a greater decrease in local trapping of 13NH3, than 18FDG within the infarct, probably because of increased anaerobic glycolysis. Otherwise, 18FDG was a more sensitive indicator of cerebral dysfunction than was 13NH3. Hypometabolism, due to deactivation or minimal damage, was demonstrated with the 18FDG scan in deep structures and broad zones of cerebral cortex that appeared normal on x-ray computed tomography and technetium 99m pertechnetate scans. In its present state of development, the 18FDG ECT method should aid in defining the location and extent of altered brain in studies of disordered function after stroke. With improved knowledge of tracer behaviour in diseased brain, the method has promise for mapping the response to therapeutic intervention and increasing our understanding of how the human brain responds to stroke.