The aim of this work was to study the feasibility and reproducibility of in vivo measurement of benzodiazepine receptors with single photon emission computerized tomography (SPECT) in the baboon brain. Arterial and brain regional activities were measured for 420 min in three baboons after single bolus injection of the benzodiazepine antagonist [123I]iomazenil. Data were fit to a three-compartment model to derive the regional binding potential (BP), which corresponds to the product of the receptor density, (Bmax) and affinity (1/KD). Regional BP values (from 114 in striatum to 241 in occipital) were in good agreement with values predicted from in vitro studies. Constraining the regional volume of distribution of the nondisplaceable compartment to the value measured during tracer constant infusion experiments in baboons (Laruelle et al., 1993) improved the identifiability of the rate constants. Each experiment was repeated to investigate the reproducibility of the measurement. The regional average reproducibility was 10 +/- 5%, expressed as coefficient of variation (CV). Results of equilibrium analysis at peak uptake were in good agreement with results of kinetic analysis. Empirical counts ratio methods were found to be poorly sensitive to benzodiazepine receptor density. These studies suggest the feasibility of quantitative measurement of benzodiazepine receptors by kinetic analysis of SPECT data and the inadequacy of empirical methods of analysis, such as counts ratios, to evaluate differences in receptor density.