Objective To investigate the usual dose and safety of rivaroxiban in super-elderly patients with non-valvular atrial fibrillation (NVAF). Methods A total of 150 super-elderly NVAF patients were enrolled as subjects, who were hospitalized in the Department of Cardiology, Air Force Medical Center from August 2017 to April 2021. The initial oral dose of rivaroxiban was determined according to the HAS-BLED score, and the patients were divided into low-dose group (10mg/d; n=78))and medium-dose group (15mg/d; n=72). Routine blood test results, coagulation, liver functions, and kidney functions were monitored before and after the treatment. The patients were followed up for 12 months for bleeding events, incidence of adverse reactions, and incidents of new ischemic strokes. SPSS statistics 19.0 was used for data analysis. Depending on data types, t test, analysis of variance (ANOVA) or χ² test was employed for intergroup comparison. Results Five patients died of severe pneumonia in each group during one-year follow-up, the difference being not statistically significant (P>0.05). Comparison between the two groups at 1,3, 6, and 12 months found no statistically significant differences (P>0.05) in hemoblobin, platelet counts, plasma prothrombin time (PT), activated partial thromboplastin time (APTT) , fibrinogen (FIB), D-dimer (DD), liver functions, and kidney functions. Recurrence of acute cerebral infarction and severe bleeding was not recorded. Symptoms with mild bleeding disappeared after adjusting dose. Conclusion Rivaroxaban (10 mg/d) is effective and safe for the super-elderly NVAF patients.