Cortical and subcortical somatosensory evoked potentials (SSEPs) were noninvasively monitored in 191 surgical procedures involving the cervical spine. In nine patients in the poorest neurological condition, SSEPs could not be monitored. Lower limb SSEPs were often too degraded to be useful. Upper limb responses were reliably recorded in 182 procedures, with a sensitivity of 99% and a specificity of 27% in 10 patients who developed neurological signs postsurgery. The aim of monitoring was to detect changes in spinal cord function at a time when neurological deterioration could be prevented or reversed, and these studies alerted the authors to certain clinical and SSEP risk factors associated with deterioration. Clinical and operative risk factors were: 1) poor pre-operative neurological function (one-third of Ranawat Class IIIb patients deteriorated); 2) use of instrumentation (the risk doubled in preoperatively unimpaired patients); 3) upper cervical and clival surgery (the risk tripled); and 4) and multisegmental surgery (increased risk with each additional level). There were SSEP changes in 33 patients. Fifty percent of patients with a complete loss had neurological damage, unlike those who had incomplete loss or whose electrical changes had recovered by the end of surgery. In the authors' view these "false positives" may represent real physiological changes, the effects of which might have been minimized by an alteration in the surgeon's response as a result of the warning. Although these initial studies have made this surgical team more alert to potential problems, the role of intraoperative SSEP monitoring is still being debated.