Objective:To evaluate Wells, Kahn, St.Andréand Constans scores for the prediction of deep venous thrombosis in Chinese patients. Methods:One hundred and seventy-two patients, prospectively, blinded referred for evaluation with four clinical-score systems for suspected deep venous thrombosis, were examined by ultrasonography. Sensitivity, specificity, positive predictive value, nega-tive predictive value and receiver operation curves were calculated for four clinical scores. The difference between areas of the ROC curve for each of the scores was compared with others and reference line. Results:Forty-six of 172 patients had deep venous throm-bosis proven by sonography. The sensitivity, specificity, positive predictive value and negative predictive value for Wells score was 91.3%, 57.1%, 43.8% and 94.7%, respectively, for Kahn score; 65.2%, 71.4%, 45.5% and 84.9%, respectively, for St.Andréscore; 63%, 38.9%, 27.4% and 74.2% respectively, for Constans score; 95.7%, 34.9%, 34.9% and 95.7% respectively. Area under ROV curve of Constans score was 0.814, which was similar to that of Wells score, then followed by Kahn score and that of St.Andréscore was no difference with the reference line. Conclusion:Based on the results of our study, the sensitivity, negative prediction value and area under ROC curve are larger for Constans score and Wells score in Chinese hospitalized patients than that of Kahn score or St.Andréscore. Considering the aim of the clinical assessment, Constans score and Wells score are more efficient for Chinese hospitalized patients.