BACKGROUND: The antiviral therapy has been considered as an ordinary intervention for COVID-19 patients. However, the effectiveness of antiviral therapy is uncertain. This study was designed to determine the association between the antiviral therapy and in-hospital mortality among severe COVID-19 patients. METHODS: This study enrolled severe COVID-19 patients admitted to four designated hospitals in Wuhan, China. The use of antiviral treatments, demographics, laboratory variables, co-morbidities, complications, and other treatments were compared between survival and fatal cases. The association between antiviral agents and in-hospital mortality were analyzed. RESULTS: In total, 109 severe COVID-19 patients (mean age 65.43) were enrolled for analysis, among which, 61 (56.0%) patients were discharged alive, and 48 (44.0%) died during hospitalization. We found no association between lopinavir/ritonavir (LPV/r) treatment and the in-hospital mortality (odds ratio (OR) = 0.195, 95% confidence interval (CI) = 0.023-1.679). Besides, ribavirin (OR = 0.738, 95% CI = 0.344-1.582), oseltamivir (OR = 0.765, 95% CI = 0.349-1.636), and interferon-alpha (IFN-α) (OR = 0.371, 95% CI = 0.112-1.236) were not associated with the in-hospital mortality. However, arbidol monotherapy (OR = 5.027, 95% CI = 1.795-14.074) or the combination of arbidol and oseltamivir (OR = 5.900, 95% CI = 1.190-29.247) was associated with an increased in-hospital mortality. In addition, the multiple logistic regression identified a significant association between the use of arbidol and the in-hospital mortality (adjusted OR = 4.195, 95% CI = 1.221-14.408). CONCLUSIONS: Our findings indicated that LPV/r, IFN-α, ribavirin, or oseltamivir have no beneficial effects on the prognosis of severe COVID-19 patients, whereas the use of arbidol is associated with increased in-hospital mortality.