The purpose of this experimental study was to examine the differences between peripheral nerve stimulation and direct spinal stimulation in generating cortical somatosensory-evoked potential (SEP) responses for monitoring spinal cord ischemia during thoracic aorta cross-clamping. Adult mongrel dogs (n = 6) were placed under general anesthesia and a left thoracotomy was performed. A conventional stimulating electrode was placed over the posterior tibial nerve (PN-SEP), and a special bipolar electrode was placed epidurally over the spinal cord at L1-2 (SC-SEP). The aorta was cross-clamped proximal to the left subclavian artery. Stimulations were alternately performed through both electrodes, and SEP responses were continuously monitored. The cross-clamp was released after one hour and the animal was observed for another hour prior to sacrifice. Excellent SEPs were obtained with six stimuli over 3 sec via the SC-SEP stimulus in contrast to the 200 stimulations over 90 sec required for the PN-SEP stimulus. Aortic cross-clamping resulted in a significantly longer mean time to loss of SEPs for SC-SEP (mean +/- SEM, 13.7 +/- 1.0 min for SC-SEP vs 11.3 +/- 0.7 min for PN-SEP, P less than 0.05). Likewise, unclamping of the aorta consistently resulted in a shorter mean time to return of SEPs for SC-SEP compared with PN-SEP. These data indicate that direct epidural stimulation for evoked cortical responses is a more sensitive means of determining the adequacy of posterior spinal cord blood flow as reflected by posterior spinal cord function.(ABSTRACT TRUNCATED AT 250 WORDS)