AIM: To estimate the prognosis of patients with liver failure using a scoring model of severe viral hepatitis (SMSVH) and a model of end stage liver disease (MELD)to provide a scientific basis for clinical decision of treatment.METHODS: One hundred and twenty patients with liver failure due to severe viral hepatitis were investigated with SMSVH established. Patients with acute, subacute,and chronic liver failure were 40, 46 and 34, respectively.The follow-up time was 6 mo. The survival rates of patients with liver failure in 2 wk, 4 wk, 3 mo and 6 mo were estimated with Kaplan-Meier method. Comparison between SMSVH and MELD was made using ROC statistic analysis.RESULTS: The survival curves of group A (at low risk,SMSVH score ≤ 4) and group B (at high risk, SMSVH score ≥ 5) were significantly different (The 4-wk, 3-mo, 6-mo survival rates were 94.59%, 54.05%, 43.24% in group A,and 51.81%, 20.48%, 12.05% in group B, respectively,P ＜ 0.001). The survival curves of group C (SMSVH scores unchanged or increased), group D (SMSVH scores decreased by 1) and group E (SMSVH scores decreased by 2 or more) were significantly different .The survival rates of groups C, D and E were 66.15%, 100%, 100% in 2-wk; 40.0%, 91.18%, 100% in 4-wk; 0%, 58.82%,80.95% in 3-mo and 0%, 38.24%, 61.90% in 6-mo,respectively, P ＜ 0.001). The area under the ROC curve (AUC) of SMSVH scores at baseline and after 2 wk of therapy was significantly higher than that under the ROC curve of MELD scores (0.804 and 0.934 vs 0.689, P ＜0.001).CONCLUSION: SMSVH is superior to MELD in the estimation of the prognosis of patients with severe viral hepatitis within 6 mo. SMSVH may be regarded as a criterion for estimation of the efficacy of medical treatment and the decision of clinical treatment.