In this study cerebral blood flow (CBF) was simultaneously measured with the Kety-Schmidt method and the intravenous 133Xe clearance technique. CBF, cerebral metabolic rate of oxygen (CMRO2), and CO2 reactivity of CBF were compared under fentanyl-midazolam anaesthesia and varying paCO2 levels. METHODS Thirteen male patients were studied before they underwent coronary artery bypass surgery. For measurement of CBF with the Kety-Schmidt inert gas saturation technique, argon was used as indicator instead of nitrous oxide, because argon is less soluble in water and lipid such that arterial and hence organ saturation is attained earlier. Wash-in periods of 10 min were used for all measurements. For measurement of CBF with the intravenous xenon method 10 scintillation detectors placed lateral to the skull and an air detector for calculation of tracer recirculation were used. 10-15 mCi of 133Xe dissolved in physiological saline was injected via a peripheral i.v. cannula. For comparison with the Kety-Schmidt technique CBF15-values representing the flow of the grey and white matter were chosen. CBF was measured simultaneously with both methods under normocapnic (paCO2 43 +/- 3 mmHg), hypocapnic (paCO2 31 +/- 3 mmHg), and under hypercapnic (paCO2 54 +/- 4 mmHg) conditions. RESULTS All CBF15 values obtained with the intravenous xenon method were significantly lower than the corresponding CBF-values measured with the Kety-Schmidt technique: by 36% under normocapnic, 23% under hypocapnic, and 39% under hypercapnic conditions, respectively. Hence, CMRO2 values calculated from CBF values obtained with the xenon method were reduced to about the same degree as those derived from CBF values measured with the Kety-Schmidt technique. There was no significant correlation between the CBF values of either method (y = 1.82x-8.58,r = 0.76 P = 0.357). Non-linear curve-fitting procedures yielded exponential CBF-paCO2 relationships for both methods, although the relative carbon dioxide reactivity was higher with the Kety-Schmidt technique than with the xenon method (y = 8.14 e0.039x vs y = 10,75 e0.023x). CONCLUSIONS Most probably due to contamination with radioactivity from slowly perfused extracerebral tissues the intravenous 133Xe-clearance technique underestimates CBF, CMRO2, and cerebrovascular CO2 reactivity, at least when CBF15 values are used as flow parameters.