We sequentially recorded subcortical (P14) and cortical (N20) somatosensory evoked potentials (SEPs) in 32 patients undergoing deep hypothermic circulatory arrest (CA). Under normal hemodynamic conditions, hypothermia initially produced N20 disappearance at a mean nasopharyngeal temperature of 20.4 +/- 2.6 degrees C (range 14.5 to 26.1 degrees C) and P14 disappearance at a mean of 16.9 +/- 2.0 degrees C (range 12.4 to 20.2 degrees C). On rewarming, P14 reappeared at mean temperature of 19.3 +/- 4.0 degrees C (range 13.5 to 29.2 degrees C) and N20 at a mean of 21.1 +/- 4.1 degrees C (range 14.3 to 29.6 degrees C). The delay of SEP reappearance after restoration of blood flow correlated significantly with CA duration (r = 0.74 for P14, and r = 0.62 for N20; p < 0.01). Neurological recovery was uneventful in 23 patients; 5 patients presented with neurological sequelae (minor or transient in 4; no recovery from anesthesia and death after 48 hours in 1), and 4 patients died during operation. Twenty-three of 24 surviving patients in whom P14 disappearance was the criterion that hypothermia was deep enough to perform CA (duration: 17 to 94 min) had a normal neurological outcome. By contrast, all surviving patients in whom cortical SEPs disappeared at higher temperatures presented neurological sequelae. In conclusion, the neurophysiological monitoring of brain stem activity, as specifically provided by SEPs, enables determination of the optimal temperature for CA, and demonstrates superiority of SEP monitoring over the use of EEG.