Positron emission tomography (PET) with 18F-fluoro-2-deoxy-D-glucose (FDG) is frequently used to study the metabolic correlates of movement and mental disorders. These studies generally focus on changes in the frontal cortex and the basal ganglia. The reproducibility of glucose metabolism estimates in these structures was tested in 13 normal subjects studied at rest using a standard and simple protocol. A reproducible dorsoventral metabolic gradient was demonstrated in the frontal cortex. Such a gradient was not present in the basal ganglia when the upper region of interest in the caudate nucleus, where the lower metabolic rate of glucose was probably attributable to partial volume effects, was not considered. Absolute values of glucose metabolic rates varied by 6.4 to 12.5% in the frontal cortex and by 6.8 to 14.7% in the basal ganglia. Variations in normalized values in the basal ganglia ranged from 4.0 to 8.6%. The number of subjects required to detect statistical differences in group comparison or in test-retest studies was calculated for different anticipated levels of change. With the variability detected in this experiment, less than 10 subjects were expected to be sufficient to detect a 15% change in most regions and in both types of studies.