Summary: Twenty-seven in-patients with hemiplegia following brain injury were studied by using upper extremity median nerve somatosensory evoked patentials (SVEP), Brunnstrom assessment in hemiplegic hand and assessment of the patients’ activities of daily lioing (ADL) (Barthel index). The upper extremity median nerve SEP on the affected and normal sides was determined. By using Kovindha standard, upper extremity median nerve SEP was graded in accordance with N20. The correlation between the differences of SEP N20 amplitude and the latencies on the both sides and the Barthel index scores was analyzed. A Spearman correlation analysis was made between the median nerve SEP N20 grades and Brunnstrom stages in hand or ADL on the affected side. The results showed that upper extremity median nerve SEP grades were positively correlated with those of the Brunnstrom stages in hand (r 1=0.6925, P 1<0.01). The correlation coefficient between SEP N20 grades and patients’ ADL grades was r 2=0.5015, P 2<0 01. It was concluded that upper extremity median nerve SEP could be used as a sensitive electrophysiological predictor to clinically assess hemiplegic hand function. SEP N20 might play a role in predicting the ADL of the patients with hemiplegia to some extent, but could not be used as a sensitive predictor to directly observe and predict the ADL of the patients. Summary: Twenty-seven in-patients with hemiplegia following brain injury were studied by using upper extremity median nerve somatosensory evoked patentials (SVEP), Brunnstrom assessment in hemiplegic hand and assessment of the patients’ activities of daily lioing (ADL) (Barthel index) The upper extremity median nerve SEP on the affected and normal sides was determined. By using Kovindha standard, upper extremity median nerve SEP was graded in accordance with N20. The correlation between the differences of SEP N20 amplitude and the latencies on the both sides and the Spearman correlation score was made between the median nerve SEP N20 grades and Brunnstrom stages in hand or ADL on the affected side. The results showed that upper extremity median nerve SEP grades were positively correlated with those of the Brunnstrom stages in hand (r 1 = 0.6925, P 1 <0.01). The correlation coefficient between SEP N20 grades and patients’ ADL grades was r 2 = 0.50 15, P 2 <0 01. It was originally that upper extremity median nerve SEP could be used as a sensitive electrophysiological predictor to clinically assess hemiplegic hand function. SEP N20 might play a role in predicting the ADL of the patients with hemiplegia to some extent , but could not be used as a sensitive predictor to directly observe and predict the ADL of the patients.