Objective To investigate the effectiveness and the best assessment time of the short-latency somatosensory evoked potential (SLSEP) and brainstem auditory-evoked potential (BAEP) in the prognosis prediction of patients with severe cerebrovascular disease. Methods A prospective trial was conducted. The patients with severe cerebrovascular disease and Glasgow coma scale (GCS) ≤ 8 and admitted to the neurological intensive care unit (NICU) of Armed Police Logistics College Affiliated Brain Hospital from December 2014 to May 2015 were enrolled. The patients received SLSEP and BAEP nerve electrophysiological examinations within 24 hours and on 3, 7, 15 days after admission respectively and were graded according to Cant method. GCS was evaluated within 24 hours and on 15 days after admission. The prognosis was evaluated by Glasgow outcome scale (GOS) at six months after the onset of the disease. At different time windows after the onset of the disease, the correlations between different predictive indexes (GCS, SLSEP and BAEP) and outcome (GOS) were analyzed using spearman rank correlation; in the mean time, the efficacy for predicting the prognosis by single index or combined indexes was compared by receiver operator characteristic (ROC) curve. Results Seventy-eight patients were enrolled [men 46, women 32, age range (60.79±12.50) years old]. There were 78, 64, 44 and 19 patients observed at 24 hours and on 3, 7, 15 days after admission because the short-term death of some patients. The graded abnormal rate of SLSEP was 75.64%, 82.81%, 79.55% and 73.98% respectively; and the graded abnormal rate of BAEP was 82.05%, 84.38%, 85.94% and 73.68% respectively. ① Correlation analysis: all the predictors were correlated with GOS within 24 hours and on 3, 7, 15 days after admission, and SLSEP and BAEP grading were moderately correlated with GOS (0.4≤|R| < 0.7). ② The accuracy of the predicting prognosis: the area under the curve (AUC) of GCS on 15 days after admission [AUC = 0.772, 95% confidence interval (95%CI) = 0.561-0.984, P = 0.045] was the maximum when predicting survival. AUC of SLSEP (AUC = 0.825, 95%CI = 0.695-0.955, P = 0.000) and BAEP (AUC = 0.786, 95%CI = 0.646-0.927, P = 0.002) were the maximum on 7 days after admission when predicting death. ③ The effectiveness of the prognosis prediction: the sensitivity of SLSEP grading and BAEP grading were 92.6% and 96.3% respectively, while the sensitivity, specificity and accuracy of SLSEP and BAEP combined prediction were 100% on 7 days after admission. The specificity of GCS was 100% on 15 days after admission. Conclusions SLSEP and BAEP have more close correlation with prognosis compared with the GCS; Continuous dynamic combined evaluation of SLSEP and BAEP has important clinical value for patients with severe cerebrovascular disease possess in the prognosis assessment, the accuracy and the effectiveness of SLSEP and BAEP combined prediction were higher on 7 days especially.