Cerebral blood flow (radioactive microspheres), intracranial pressure (subdural bolt), and retinal histopathology were examined in 20 dogs undergoing 150 minutes of hypothermic (28 degrees C) cardiopulmonary bypass to compare alpha-stat (arterial carbon dioxide tension, 40 +/- 1 mm Hg; n = 10) and pH-stat (arterial carbon dioxide tension, 61 +/- 1 mm Hg; n = 10) techniques of arterial carbon dioxide tension management. Pump flow (80 mL.kg-1.min-1), mean aortic pressure (78 +/- 2 mm Hg), and hemoglobin level (87 +/- 3 g/L [8.7 +/- 0.3 g/dL]) were maintained constant. During bypass, intracranial pressure progressively increased in the alpha-stat group from 6.0 +/- 1.0 to 13.9 +/- 1.8 mm Hg (p less than 0.05) and in the pH-stat group from 7.7 +/- 1.1 to 14.7 +/- 1.4 mm Hg (p less than 0.05), although there was no evidence of loss of intracranial compliance or intracranial edema formation as assessed by brain water content. With cooling, cerebral blood flow decreased by 56% to 62% in the alpha-stat group (p less than 0.05) and by 48% to 56% in the pH-stat group (p less than 0.05). However, 30 minutes after rewarming to 37 degrees C, cerebral blood flow in both groups failed to increase and remained significantly depressed compared with baseline values. Both groups showed similar amounts of ischemic retinal damage, with degeneration of bipolar cells found in the inner nuclear layer in 67% of animals. We conclude that, independent of the arterial carbon dioxide tension management technique, (1) cerebral perfusion decreased comparably during prolonged hypothermic bypass, (2) intracranial pressure increases progressively, (3) ischemic damage to retinal cells occurs despite maintenance of aortic pressure and flow, and (4) a significant reduction in cerebral perfusion persists after rewarming.