In patients with cerebrovascular disease, hypercarbia may cause redistribution of regional cerebral blood flow from marginally perfused to well-perfused regions (intracerebral steal), as evidenced by regional cerebral blood flow studies during carotid endarterectomy. During hypothermic cardiopulmonary bypass, the pH-stat method of acid-base management produces relative hypercarbia. To determine whether pH-stat management produces relative hypercarbia. To determine whether pH-stat management induces intracerebral steals, we investigated nine patients with cerebrovascular disease undergoing coronary artery bypass grafting. During hypothermic cardiopulmonary bypass, arterial carbon dioxide tension was varied in random order between 40 mm Hg and 60 mm Hg (uncorrected for body temperature). Regional cerebral blood flow was measured by clearance of 133 xenon injected into the arterial inflow cannula. Nasopharyngeal temperature (26.8 degrees-28.0 degrees +/- 2.2 degrees-3.0 degrees C), perfusion flow rate (2.14-2.18 +/- 0.70-0.73 L/min/m2), mean arterial pressure (67-68 +/- 6-9 mm Hg), arterial carbon dioxide tension (302-308 +/- 109-113 mm Hg), and hematocrit (23% +/- 4%) were maintained within narrow limits in each patient during arterial carbon dioxide tension manipulation. Global mean cerebral blood flow values were similar to previously reported values in patients free of cerebrovascular disease; patients in this study averaged 15.2 +/- 2.5 ml/100 gm/min at an arterial carbon dioxide tension of 46.1 +/- 8.4 mm Hg and 25.3 +/- 6.1 ml/100 gm/min at an arterial carbon dioxide tension of 71.1 +/- 11.8 mm Hg. Carbon dioxide reactivity, defined as mean global cerebral blood flow (in ml/100 gm/min) divided by arterial carbon dioxide tension (in mm Hg), was similar in the region having the lowest regional cerebral blood flow and in the brain as a whole. No patient developed evidence of an intracerebral steal at the higher arterial carbon dioxide tension. During hypothermic cardiopulmonary bypass, higher levels of arterial carbon dioxide tension, such as those associated with the pH-stat management technique, are apparently not associated with potentially harmful redistribution of cerebral blood flow in patients with cerebrovascular disease.