Objective.Near-infrared spectroscopy (NIRS) has the potential forproviding valuable information about oxygen delivery to the brain. However,questions have been raised about the accuracy of these measurements. Thisstudy was undertaken to compare noninvasive cerebral saturation measurementsto jugular venous saturation under conditions of hypoxia and hypercapnia.Methods.Data was obtain on forty-two subjects. Cerebral oxygenationwas measured with a Somanetics INVOS 4100-SSA placed on the forehead of thesubjects. PETCO2 was controlled to approximately 2 and 7 mmHg aboveresting values and PETO2 was controlled to 80, 45, 60 and 41 mmHgconsecutively for four of five minutes each. Internal jugular blood gasmeasurements were made via a retrograde catheter. Results.Both thecerebral oximetry measured saturation (rSO2) and the jugular venoussaturation (Sj $$\bar v $$ O2) were significantly increasedby increasing the PETCO2 at all levels of hypoxia. The increase inthe rSO2 was less than the increase inSj $$\bar v$$ O2. The rSO2 had a bias of 5.2%and a precision of 10.7% compared to the measuredSj $$\bar v$$ O2. Discussion.Cerebral oxygensaturation measured by cerebral oximetry compares well to the measuredSj $$\bar v$$ O2 in normal subjects, despite multiplephysiological reasons for differences. The closer relationship ofSj $$\bar v$$ O2 to rSO2 than SaO2under the conditions of these experiments indicates that the measurementreflects primarily intracranial saturation. However, outcome studies underclinical conditions are needed to determine the clinical utility of cerebraloximetry.