Hyperoxic therapy for cerebral ischemia reduces cerebral blood flow (CBF) principally from the vasoconstrictive effect of oxygen on cerebral arterioles. Based on a recent study in normal volunteers, we now claim that the vasodilatory effect of carbon dioxide predominates when 5% CO(2) is added to inhaled oxygen (the mixture known as carbogen). In the present study, we measured CBF by positron emission tomography (PET) during inhalation of test gases (O(2), carbogen, and atmospheric air) in healthy volunteers (n = 10) and in patients with occlusive carotid artery disease (n = 6). Statistical comparisons by an additive ANOVA model showed that carbogen significantly increased CBF by 7.51 + or - 1.62 ml/100 g/min while oxygen tended to reduce it by -3.22 + or - 1.62 ml/100 g/min. A separate analysis of the hemisphere contralateral to the hypoperfused hemisphere showed that carbogen significantly increased CBF by 8.90 + or - 2.81 ml/100 g/min whereas oxygen inhalation produced no reliable change in CBF (-1.15 + or - 2.81 ml/100 g/min). In both patients and controls, carbogen was as efficient as oxygen in increasing Sa(O2) or PaO(2) values. The study demonstrates that concomitant increases of CBF and Sa(O2) are readily obtained with carbogen, while oxygen increases only Sa(O2). Thus, carbogen improves oxygen transport to brain tissue more efficiently than oxygen alone. Further studies with more subjects are, however, needed to investigate the applicability of carbogen for long-term inhalation and to assess its therapeutic benefits in acute stroke patients.