Although hypothermic circulatory arrest and low-flow cardiopulmonary bypass are routinely used for surgical correction of congenital cardiac anomalies, use of long durations of arrest, often required for more complex repairs, raises serious concerns about cerebral safety. Searching for an intraoperative assessment that can reliably predict cerebral injury, we have found an excellent correlation between changes in quantitative electroencephalography intraoperatively and immediately postoperatively after prolonged hypothermic arrest, and neurologic and behavioral evidence of cerebral injury. After epidural placement of four recording electroencephalographic electrodes and baseline neurologic/behavioral and electroencephalographic assessment, 32 puppies were randomly assigned to one of four groups: hypothermic controls in which cooling to 18 degrees C was followed immediately by rewarming, 30 minutes of hypothermic circulatory arrest at 18 degrees C, 90 minutes of arrest at 18 degrees C, and 90 minutes of low-flow cardiopulmonary bypass at 25 ml/kg per minute at 18 degrees C. An electroencephalogram was recorded at baseline, after cooling, during rewarming, and at 2, 4, and 8 hours after the start of rewarming, as well as before the operation and 1 week after the operation. Postoperative neurologic and behavioral outcome was assessed 24 hours after cardiopulmonary bypass and daily for 1 week by means of a graded scale in which 0 is normal and 12 and 13 indicate severe neurologic injury (coma and death). Thirty animals survived the experimental protocol: two animals in the 90-minute hypothermic arrest group died before neurologic evaluation could be completed, and the remainder exhibited various degrees of neurologic and behavioral impairment, more severe on day 1 than on day 6. No animal in the remaining groups had a significant neurologic deficit. Quantitative electroencephalographic analysis shows marked differences between the 90-minute arrest group and the controls in the percent electroencephalographic silence during rewarming and at 2 hours, and in the percent recovery of baseline power at 2, 4, and 8 hours. At 2 hours after the start of rewarming, a correlation between electroencephalographic amplitude and neurologic/behavioral score on day 1 was carried out, which predicts with great certainty (p < 0.00001) that if electroencephalographic power at this time is less than 500 microV2, overt neurologic injury will subsequently become apparent. In addition, a significant shift from higher to lower frequency in the day 6 postoperative electroencephalogram compared with baseline occurs only in the 90-minute arrest group.(ABSTRACT TRUNCATED AT 400 WORDS)