The usefulness of spinal motor evoked potential by transcranial stimulation of the motor cortex (MEPT) in detecting spinal ischemia and predicting postoperative neurological dysfunction was evaluated using a model of spinal ischemia. Group 1 was comprised of 11 dogs used for measuring the basic wave form of spinal MEPT. The normal spinal MEPT response curve consists of two major peaks: peak I and peak II. The latency of peak I and peak II at T13-L1 was 6.0 +/- 0.6 and 7.1 +/- 0.6 msec, and the amplitude, 3.3 +/- 1.6 and 6.1 +/- 2.6 microV, respectively. Group 2 was comprised of six animals subjected to spinal ischemia, in which a time-related deterioration of the MEPT as well as evoked spinal cord potential (ESP) was demonstrated. The time taken until the loss of peak I and peak II was 19.2 +/- 5.3 and 21.7 +/- 6.2 min, respectively, while the time taken until the loss of ESP was 36.7 +/- 14.0 min. In group 3, comprised of seven animals, the aorta was unclamped and the animals were allowed to recover when the spinal MEPT had disappeared. Four had paraparesis immediately after the operation, two had a normal gait, one died, and one developed spastic paraplegia after 24 h. We concluded that the change in spinal MEPT during spinal ischemia occurred earlier than the change in ESP, and that the loss of MEPT suggested irreversible spinal cord damage.