BACKGROUND: Near infrared spectroscopy (NIRS) to monitor regional cerebral oxygen saturation (rSO2) is a noninvasive and simple modal ity in clinical use. The ability of rSO2 as an index of cerebral oxygenation has been well demonstrated. However, the reliability of rSO2 to reflect the changes of cerebral vascular reactivity in the changes of arterial partial pressure of CO2 (PaCO2) has not been established. The aim of this study was to verify the reliability of rSO2 to measure the CO2 reactivity of cerebral vasculatures. METHODS: Twenty healthy adult patients undergoing general anesthesia were enrolled in this study. Anesthesia was induced with propofol and maintained with desflurane/N2O. Respiration was mechanically controlled. The radial artery and jugular bulb were cannulated. The sensor of the NIRS was attached to the ipsilateral forehead. During normocapnia (PaCO2 40 +/- 1.3 mmHg) and hypocapnia (PaCO2 30 +/- 2.4 mmHg), blood was obtained from the radial artery and jugular bulb and analyzed. rSO2 was compared with fSO2 (estimated field oxygen satuation), and the gold standard of tissue oxygen saturation. fSO2 was calculated from the following equation: fSO2 = 0.75 SjO2 + 0.25 SaO2. RESULTS: rSO2 significantly correlated with fSO2 (P = 0.000, r2 = 0.56). A bias of - 5.8% with a precision 12.94% was found. CONCLUSIONS: We concluded that rSO2 can be a reliable predictor to measure CO2 reactivity of cerebral vasculatures during normocapnia and hypocapnia.