Background. The impact of corticosteroids on severe patients with coronavirus disease 2019 (COVID-19)/ chronic hepatitis B virus (HBV) co-infection is currently unknown. We aimed to investigate effect of corticosteroid on these subgroup patients.Methods. In this retrospective multicenter study including 5447 confirmed COVID-19 patients from Jan 1, 2020 to Apr 18, 2020, severe patients with COVID-19/HBV co-infection were identified. To minimize the bias of confounding variables on effect of corticosteroid treatment, inverse probability of treatment weighting (IPTW) based on propensity score was employed.Results. The prevalence of HBV co-infection in hospitalization COVID-19 patients was 4.1%. 105 severe patients with COVID-19/HBV co-infection were enrolled (median age 62 years, 57.1% male). Fifty-five patients received corticosteroid treatment and 50 patients did not. Corticosteroid treatment was associated with high D-dimer level, neutrophil count (all P<0.05). With IPTW analysis, corticosteroid treatment worsen acute liver injury (OR, 1.767, 95%CI, 1.018-3.065, P=0.043). Corticosteroids might delay SARS-CoV-2 viral RNA clearance (OR, 4.963, 95%CI, 2.717-9.065, P <0.001). The 28-day and in-hospital mortality were both significantly higher in corticosteroid treatment group than non-corticosteroid treatment group (OR, 8.738, 95%CI, 2.826-27.022, P <0.001; OR, 10.122, 95%CI, 3.291-31.129, P <0.001, respectively). In multivariable analysis, higher D-dimer level (>1µg/ml) (OR, 10.686, 95%CI, 2.421-47.159, P=0.002) and corticosteroid therapy (OR, 11.236, 95%CI, 1.273-99.154, P=0.029) were independently associated with 28-day mortality. Methylprednisolone dose per day and cumulative dose in non-survivors was significantly higher than in survivors.Conclusions. In severe patients with COVID-19/HBV co-infection, corticosteroid treatment may increase mortality. Therefore, corticosteroid therapy should be prescribed with caution in the subset of patients.