The aim of this study was to analyze the influence of variations in the size of regions of interest (ROIs) on values of caudate glucose consumption [regional cerebral metabolic rate of glucose (rCMRglc)] assessed by PET. In addition, we tried to establish the influence of ROI size on levels of significance assessing differences in mean caudate glucose consumption between two groups of subjects. For this purpose, rCMRglc was measured using [18F]fluorodeoxyglucose and the PC-4096 PET camera with an in-plane resolution of 7.1 mm in 12 normal subjects and 12 subjects with early Huntington disease. Caudate rCMRglc was histographically determined using 10 different ROI widths ranging from 2 mm, corresponding to the peak value of caudate rCM-Rglc, to 20 mm. The increase in ROI width from 2 to 20 mm led to a significant decrease of caudate rCMRglc by approximately 66% in the normal subjects and in the patients. The Student t value assessing the differences in mean caudate rCMRglc between the two groups decreased gradually from 5.61 for an ROI width of 2 mm to 3.78 for an ROI width of 20 mm. This corresponds to a worsening of the level of significance from 10(-5) to 10(-3), i.e., by a factor of 100. These data show that (given the resolution of presently used PET cameras) the selection of peak values of caudate rCMRglc is the best way to discriminate between groups of subjects supposed to differ with respect to caudate rCMRglc.