BACKGROUND: Selective cerebral perfusion (SCP) may enhance the neuroprotective benefits of hypothermia during aortic surgery. However, despite its widespread adoption, there is no consensus regarding optimal implementation of SCP. We used a survival porcine model to explore the physiologic characteristics and behavioral benefits of various protocols involving hypothermic circulatory arrest (HCA) and SCP. METHODS: Thirty pigs (26.3 +/- 1.4 kg), cooled to 15 degrees C on cardiopulmonary bypass, using alpha-stat pH management (mean hematocrit 30%), were randomly allocated to differing brain protection strategies: 90 minutes of HCA (group A); 30 minutes of HCA, then 60 minutes of SCP (group B); or 90 minutes of SCP (group C). Using fluorescent microspheres and sagittal sinus sampling, cerebral blood flow (CBF [mL x 100g(-1) x min(-1)]) and cerebral metabolic rate for oxygen (CMRO2 [mL x 100g(-1) x min(-1)]) were assessed at baseline, after cooling, during SCP (where applicable), and for 2 hours after cardiopulmonary bypass. Neurobehavioral scores were assessed blindly from standardized videotaped sessions for 7 days postoperatively. RESULTS: Cerebral blood flow was significantly higher (p = 0.0001) during SCP (60 and 90 minutes) if preceded by HCA. The CMRO2 was also significantly higher in group B versus group C (p = 0.016) at 60 minutes. The CMRO2 in all three groups rebounded promptly toward baseline after weaning from cardiopulmonary bypass. Postoperative neurobehavioral scores were significantly worse in group A. CONCLUSIONS: Continuous SCP provides the best brain protection overall. However, an initial period of HCA does not seem to impair late outcome; perhaps the elevated CBF and CMRO2 observed reflect a beneficial cerebral response to a recoverable insult. Clearly, 90 minutes of HCA induces permanent brain injury, even in this carefully controlled setting.