OBJECTIVE Cardiopulmonary resuscitation (CPR) generating low perfusion pressures and beginning immediately after cardiac arrest maintains cerebral ATP but not cerebral pH or arterial pH. We tested the hypothesis that preventing severe arterial acidemia prevents cerebral acidosis, whereas augmenting arterial acidemia augments cerebral acidosis. METHODS In dogs anesthetized with pentobarbital and fentanyl, cerebral pH and ATP were measured with 31P MR spectroscopy and blood flow was measured with radiolabeled microspheres. A pneumatically controlled vest was placed around the thorax, and chest compressions were begun immediately after electrically induced cardiac arrest. Cerebral perfusion pressure was maintained with epinephrine at 30 mm Hg for 90 minutes. The arterial acidemia observed during CPR was untreated in a control group, corrected to a pH of 7.3 with the use of sodium bicarbonate, or maintained below pH 6.5 with intravenous lactic acid after 14 minutes of CPR. RESULTS At 10 minutes of CPR, cerebral ATP (99 +/- 1.5%, control), blood flow (35 +/- 3 mL/min per 100 g), O2 consumption (4.0 +/- 0.2 mL/min per 100 g), and cerebral pH (7.05 +/- .03) were unchanged from prearrest values (mean +/- SEM). After 10 minutes of CPR in the control group, cerebral pH progressively fell (6.43 +/- 0.10 at 90 minutes) in parallel with cerebral venous pH. In the bicarbonate group cerebral pH was maintained higher (6.91 +/- 0.08). Cerebral blood flow, O2 consumption, and ATP were sustained near prearrest values in both groups. In the lactate group, however, the rate of decrease of cerebral pH was augmented (6.47 +/- 0.06 by 30 minutes), and cerebral blood flow and metabolism were significantly reduced. CONCLUSIONS Cerebral pH decreased in parallel with blood pH when resuscitation was started immediately upon arrest even when cerebral O2 consumption and blood flow were near normal. Although cerebral metabolism was near normal during the first hour of CPR, systemic bicarbonate administration ameliorated the cerebral acidosis. This finding indicates that the blood-brain pH gradient is important at the subnormal cerebral perfusion pressures seen in CPR.