Factors contributing to the quantification of the central benzodiazepine receptor binding potential in vivo using I-123-iomazenil and single photon emission computed tomography (SPECT) were analysed in phantom studies and in volunteers. SPECT was performed with the SME810 multidetector system. The Hoffman three-dimensional brain phantom was used to investigate linearity and reproducibility of SPECT results using different I-123 activity concentrations. Dynamic and multislice SPECT scans were performed in nine healthy volunteers between 0 and 270 min after injection of 110 MBq I-123-iomazenil. Displacement studies were performed in three volunteers using 1 mg flumazenil administered intravenously (i.v.) at 250 min postinjection (p.i.). Regions of interest (ROIs) over the cerebral cortex and various subcortical nuclei were either fixed areas based on a stereotaxic brain atlas, or manually drawn, based on 60% isocontour lines. Interobserver variability of the ROI data and ratios derived from those data were estimated for two independent observers. The brain phantom measurements showed linearity with respect to I-123 concentration and good reproducibility. The interobserver study showed a reasonable interobserver reliability for the large fixed ROIs. The displacement study showed about 50% displacement in all ROIs. Concentrations of I-123-iomazenil in the brain can be measured reliably with the SME810. Large fixed ROIs based on a stereotaxic atlas may be used reliably for quantification. The level of 4 cm above the cantomeatal line does not provide a useful reference region with a very low central benzodiazepine receptor binding potential.