OBJECTIVES: Whether combination antiviral therapy is cost-effective for patient with HCV in the real-world setting has not yet to be shown. This study is to compare the cost-effectiveness of combination antiviral therapy with no treatment in cirrhotic patients with ≥1 doses of combination prescriptions (base case), and with ≥48 weeks continuous combination prescriptions (subgroup) from a managed care organization health care system perspective.   METHODS: Using the Integrated Health Care Information Services National Managed Care Benchmark Database from January 19997-June 2007, we followed newly-diagnosed, HCV-infected patients with cirrhosis for the occurrence of end-staged liver diseases (ESLD), starting from the date of their first prescriptions. For each patient, we measured effectiveness (Ei) as time to ESLD occurrence or censored, and costs (Ci) as the sum of pharmacy, outpatient and inpatient costs incurred in the follow up (2007 US dollars). Individual patient’s net benefit was calculated as lambda times Ei minus Ci, and the net benefit regression framework was employed to determine the cost-effectiveness of antiviral therapy and the importance of covariates on the incremental net benefit of an intervention controlling for patient-related characteristics. The inverse probability of censored weighting technique was used to account for potential bias due to censored cost and effectiveness data. RESULTS: The results found that, at a number of lambda values ≥US$10,000, combination antiviral therapy was more cost effective than no treatment in base case analysis. In subgroup analysis, continuous antiviral therapy was cost effective than no treatment at lambda value ≥US$1000,000. Time of HCV diagnosis, HIV co-infection and prior gastroenterologist visits were potentially important covariates on the incremental net benefit of treatment. CONCLUSIONS: From the standpoint of cost-effectiveness, continuation of antiviral therapy should consider the variability in early viral response because treatment beyond 12 weeks is no recommended for genotype 1, HCV-infected patients when achieving sustained viral response is unlikely.