There is some controversy regarding the value of intraoperative neurophysiological monitoring in predicting postoperative neurological deficits. We discuss our experience with the use of intraoperative somatosensory evoked potentials (SSEPs) during surgery of cranial base tumors. We retrospectively reviewed all of the procedures that had been performed for the resection of cranial base tumors from July 29, 1993, through March 16, 1995. One hundred ninety-three consecutive patients had undergone a total of 244 procedures. SSEP waveforms were classified as follows: Type I, no change; Type II, change that reverts to baseline; Type III, change that does not revert to baseline; and Type IV, complete flattening of the SSEP waveform without improvement. Two patients had no waveforms from the beginning of the case (Type V) and were excluded from further analysis. New immediate postoperative neurological deficits were recorded. There were 64 male and 129 female patients, with a mean age of 46.6 years. One hundred seventy-seven patients had Type I SSEP waveforms, 13 of whom had postoperative deficits (7%). Fifty-six patients had Type II SSEPs, and nine (16%) of them had postoperative neurological deficits. Six patients had Type III SSEPs, and three had Type IV SSEPs, all of whom (100%) had postoperative deficits. There was a correlation between SSEP type and the results of the postoperative neurological examinations. The positive predictive value is 100%, and the negative predictive value is 90%. Although a change in the waveform that did not revert to baseline (Types III and IV) always predicted a postoperative deficit, a normal waveform did not always rule out postoperative deficits. Pathological abnormality, vessel encasement, vessel narrowing, degree of cavernous sinus involvement, brain stem edema, middle fossa location, final amount of resection, age, and tumor size correlated with a high predictive value of SSEP monitoring on univariate analysis (P < 0.05). None of these variables correlated significantly on multivariate analysis (P > 0.05), although brain stem edema was close (P = 0.0571). Intraoperative SSEPs have a high positive predictive value during surgery for cranial base tumors, but they do not detect all postoperative deficits.