Background Surgery is usually the treatment of choice for patients with cervical compressive myelopathy (CCM). Motor evoked potential (MEP) has proved to be helpful tool in evaluating intraoperative cervical spinal cord function change of those patients. This study aims to describe and evaluate different MEP baseline phenotypes for predicting MEP changes during CCM surgery. Methods A total of 105 consecutive CCM patients underwent posterior cervical spine decompression were prospectively collected between December 2012 and November 2016. All intraoperative MEP baselines recorded before spinal cord decompression were classified into 5 types (I to V) that were carefully designed according to the different MEP parameters. The postoperative neurologic status of each patient was assessed immediately after surgery. Results The mean intraoperative MEP changes range were 10.2% +/- 5.8, 14.7% +/- 9.2, 54.8% +/- 31.9, 74.1% +/- 24.3, and 110% +/- 40 in Type I, II, III, IV, and V, respectively. There was a significant correlation of the intraoperative MEP change rate with different MEP baseline phenotypes (r = 0.84, P < 0.01). Postoperative transient new spinal deficits were found 0/31 case in Type I, 0/21 in Type II, 1/14 in Type III, 2/24 in Type IV, and 4/15 in Type V. No permanent neurological injury was found in our cases series. Conclusions The MEP baselines categories for predicting intraoperative cervical cord function change is proposed through this work. The more serious the MEP baseline abnormality, the higher the probability of intraoperative MEP changes, which is beneficial to early warning for the cervical cord injury.