BACKGROUND. Recently retrograde cerebral perfusion (RCP) has been advocated as an alternative to complete circulatory arrest during aortic arch surgery. METHODS. In 19 baboons, we compared brain protection using hypothermic circulatory arrest or RCP. Animals were placed on cardiopulmonary bypass, cooled to 18 degrees C, underwent 1 hour of circulatory arrest or RCP, and were reperfused for 3 hours. Biochemical variables, cerebral blood flow (colored microsphere technique), and brain histology were assessed. RESULTS. Release of the brain-specific ischemic marker CK-BB was similar in both groups (peak values, 123 +/- 97 U/L in the circulatory arrest group and 164 +/- 88 U/L in the RCP group; p > 0.05), as were the arteriovenous differences in glucose uptake and lactate production (p > 0.05). During RCP, significant brain flow could not be detected (0.5 +/- 0.5 mL.min-1 x 100 g-1). About 90% of the blood was shunted to the inferior caval vein, and an equilibrium in circulating microspheres was found between RCP inflow and caval vein outflow. Less than 1% of the RCP inflow returned to the aortic arch. Histologic signs of brain damage were minimal in both groups, although slightly more glial edema was found in the RCP group. CONCLUSIONS. These data suggest that in nonhuman primates, retrograde cerebral perfusion does not perfuse the brain because of venovenous shunting.