METHODS Acute spinal cord injury was induced by a clip compression model in rats to approximate spinal cord injury encountered in spinal surgery. Spinal somatosensory-evoked potential neuromonitoring was used to study the electrophysiologic change. OBJECTIVE To compare and correlate changes in evoked potential after acute compression at different core temperatures with postoperative neurologic function and histologic change, to evaluate current intraoperative neuromonitoring warning criteria for neural damage, and to confirm the protective effect of hypothermia in acute spinal cord compression injury by electrophysiologic, histologic, and clinical observation. BACKGROUND With the increase in aggressive correction of spinal deformities, and the invasiveness of surgical instruments, the incidence of neurologic complication appears to have increased despite the availability of sensitive intraoperative neuromonitoring techniques designed to alert surgeons to impending neural damage. Many reasons have been given for the frequent failures of neuromonitoring, but the influence of temperature-a very important and frequently encountered factor-on evoked potential has not been well documented. Specifically, decrease in amplitude and elongation of latency seem not to have been sufficiently taken into account when intraoperative neuromonitoring levels were interpreted and when acceptable intraoperative warning criteria were determined. METHODS Experimental acute spinal cord injury was induced in rats by clip compression for two different intervals and at three different core temperatures. Spinal somatosensory-evoked potential, elicited by stimulating the median nerve and recorded from the cervical interspinous C2-C3, was monitored immediately before and after compression, and at 15-minute intervals for 1 hour. RESULTS Spinal somatosensory-evoked potential change is almost parallel to temperature-based amplitude reduction and latency elongation. Significant neurologic damage induced by acute compression of the cervical spinal cord produced a degree of effect on the amplitude of spinal somatosensory-evoked potential in normothermic conditions that differed from the effect in moderately hypothermic conditions. Using the same electromonitoring criteria,moderately hypothermic groups showed a significantly higher false-negative rate statistically (35%) than normothermic groups (10%). CONCLUSIONS Systemic cooling may protect against the detrimental effects of aggressive spinal surgical procedures. There is still not enough published information available to establish statistically and ethically acceptable intraoperative neuromonitoring warning and intervention criteria conclusively. Therefore, an urgent need exists for further investigation. Although a reduction of more than 50% in evoked potential still seems acceptable as an indicator of impending neural function loss, maintenance of more than 50% of baseline evoked potential is no guarantee of normal postoperative neural function, especially at lower than normal temperatures.