The non-invasive 133Xe inhalation method was modified to measure regional cerebral blood flow (rCBF) of both hemispheres and brain stem-cerebellar (BSC) regions in baboons to serve as a model for clarifying and validating human studies. Measurements were made in the alert, unanesthetized state at rest and during activation with spontaneous respiration in a primate chair or during sedation or anesthesia with mechanical respiration for direct comparison with the intra-arterial 19SXe injection method. Satisfactory gamma counts recorded from the brain stem by suitably directed and collimated probes overlying the suboccipital region were confirmed in vivo and by mapping isocount curves from brain tissue immediately postmortem. There was good correlation between measurements of fast flow (Fg) and slow flow (Fw) in hemispheric and BSC regions obtained by inhalation and injection methods. Extracerebral contamination of Fg clearance curves after inhalation of 133Xe was minimal. ISI2 (inhalation) correlated best with CBF10 (injection). Reproducibility of the inhalation values for Fg was r = 0.95 (p < 0.001). CO2 responsiveness measured by inhalation and injection methods showed good agreement. Fg values during wakefulness were 90.0 ± 4.0 ml/lOOg brain/min for the hemispheres and 90.0 ± 2.6 for BSC regions. Both sedation and anesthesia decreased Fg more in BSC regions than in the hemispheres. Unilateral sensory-motor activation or intravenous injection of pentylenetetrazol increased Fg both in hemispheric and BSC regions. Results appear to validate rCBF measurements by >9SXe inhalation method in humans. The method appears to be applicable to animal models where painless, serial measurements of organ blood flow without anesthesia are required.