Cerebral glucose metabolic rates were measured in 23 normal volunteers by studying the uptake of 18F-deoxyglucose with positron emission tomography (PET). These values were correlated against a brain-size index obtained from the PET images by measuring the lateral and anteroposterior brain diameters. There was a significant negative correlation (p less than 0.001), which corresponds, within the statistical uncertainty, to an inverse proportionality of glucose metabolic rate and brain volume. The variation with brain size accounted for approximately one-half of the total intersubject variance. Measurements on phantoms of different sizes and analyses of instrumental errors failed to uncover any artifactual reason for the correlation.