In our retrospective cohort study, we aim to explore whether Azvudine modifies the risk of death in COVID-19 patients. It was conducted on the medical records of patients, consecutively admitted for COVID-19 pneumonia to two hospitals in Chongqing, China. Based on Azvudine treatment exposure, the patients were divided into Azvudine group and non-Azvudine group. We used 1:2 ratio propensity score matching (PSM) in our study to adjust for confounding factors and differences between Azvudine and non-Azvudine groups. There were 1072 patients included in our original cohort. With 1:2 ratio PSM, the Azvudine group included 195 patients and non-Azvudine group included 390 patients. The results showed that Azvudine treatment was associated with improved in-hospital mortality in overall population (OR 0.375, 95% CI 0.225-0.623, P < 0.001), severe subgroup (OR 0.239, 95% CI 0.107-0.535, P = 0.001), critical subgroup (OR 0.091, 95% CI 0.011-0.769, P = 0.028) in matched cohort with univariate analysis. And there was a significantly lower in-hospital mortality in overall population (11% vs. 24%, P＜0.001), severe sub-group (10% vs. 32%, P < 0.001) and critical sub-group (5% vs. 34%, P = 0.017) in matched cohort. These results suggest Azvudine can reduce in-hospital mortality in overall COVID-19 patients, severe, and critical subgroup population.