"Delayed" single-photon emission tomograpic (SPET) images after an intravenous bolus injection of iodine-123 iomazenil have been used as a relative map of benzodiazepine receptor binding. We determined the optimal scan time for obtaining such a map and assessed the errors of the map. SPET and blood data from six healthy volunteers and five patients were used. A three-compartment kinetic model was employed in simulation studies and analyses of actual data. The simulation studies suggested that, in the normal brain, the scan time at which a single SPET image best represented the relative receptor binding was 3.0-3.5 h post-injection. This finding was supported by actual data from the volunteers. The simulation studies also suggested that the optimal scan time was not greatly changed by the variability of the input functions, and that the error in the SPET image contrast in the vicinity of the optimal scan time was not increased by changes in the tracer kinetics in the entire brain. The SPET image contrast in the patients at 3.0 h post-injection agreed well with the reference receptor binding estimated by kinetic analysis, with a mean error of 3.6%. These findings support the use of a single SPET image after bolus injection of [123I]iomazenil as a relative map of benzodiazepine receptor binding. For this purpose, a SPET scan time of 3.0-3.5 h post-injection is recommended.