Introduction: The ideal hematocrit (HCT) level during hypothermic selective cerebral perfusion (SCP) - to ensure adequate oxygen delivery without excessive perfusion - has not yet been determined. Methods: Twenty pigs (26.0+/-2.6kg) were randomized to low or high HCT management. The cardiopulmonary bypass (CPB) circuit was primed with crystalloid in the low HCT group (21+/-1%), and with donor blood in the high HCT group (30+/-1%). Pigs were cooled to 20 degrees C and SCP was carried out for 90min. During rewarming, whole blood was added in the low HCT group and crystalloid in the high HCT group to produce equivalent HCT levels by the end of the procedure. Using fluorescent microspheres and sagittal sinus sampling, cerebral blood flow (CBF) and oxygen metabolism (CMRO(2)) were assessed at baseline, after cooling, at two points during SCP (30 and 90min), and at 15min and 2h post-CPB. In addition, a range of physiological and metabolic parameters, including intracranial pressure (ICP), were recorded throughout the procedure. The animals' behavior was videotaped and assessed blindly for 7 days postoperatively (maximum score=5). Results: HCT levels were equivalent at baseline, 2h post-CPB, and at sacrifice, but significantly different (p<0.0001) during cooling and SCP. Mean arterial pressure, pH and pCO(2), and CMRO(2) were equivalent between groups throughout. ICP was similar in the two groups throughout cooling, SCP, and rewarming, but was significantly higher in the low HCT animals after the termination of CPB. CBF was similar at baseline, but thereafter markedly higher in the low HCT group. Neurobehavioral performance was significantly better in the high HCT animals (median score 3.5 vs 4.5 on day 3, and 4.5 vs 4.75 on day 7, p=0.003). Conclusions: Higher HCT levels for SCP produced a significantly superior functional outcome, suggesting that the higher CBF with a lower HCT may be injurious, possibly because of an increased embolic load.