Laboratory studies demonstrate that mild degrees of brain cooling (2 degrees C to 5 degrees C) confer substantial protection from ischemic brain injury, and that mild elevation of brain temperature can be markedly deleterious. During hypothermic cardiopulmonary bypass (CPB) patients are made hypothermic and then rewarmed at a time when they are exposed to neurological insults. Our studies show that during rewarming, peak brain temperatures near 39 degrees C often are achieved inadvertently. We hypothesize that maintaining brain temperature < or = 34 degrees C during and after CPB will reduce the incidence of postoperative neuropsychological deficits. We present safety data from a study of 30 patients assigned either to conventional hypothermic CPB with rewarming or a protocol where brain temperature is raised only to 34 degrees C at the time of separation from CPB. There was no difference in bleeding, cardiac morbidity, or time to extubation between groups. We designed a neuropsychological test battery to detect postoperative neuropsychological deficits and tested its usefulness in a preliminary sample of 15 patients undergoing hypothermic CPB. We found patient acceptability and compliance were good. Sensitivity also seemed adequate in that 30% of patients were identified as having deteriorated at 1 week postoperatively compared to preoperatively, a result similar to that reported by others. Clinical trials of the efficacy of mild hypothermia in modulating brain injury in humans are needed before techniques of CPB can be designed to optimize neuroprotection.