Edoxaban is a once-daily (QD) oral, direct factor Xa inhibitor in clinical development for the prevention of stroke in patients with non-valvular atrial fibrillation (NVAF). The aim of this study was to evaluate the safety of edoxaban in Japanese patients with NVAF. A total of 536 NVAF patients (CHADS2 ≥1) were randomized to receive double-blinded edoxaban 30, 45, or 60 mg QD or open-label warfarin (international normalized ratio [INR] 2.0-3.0 for age <70 years; 1.6-2.6 for age ≥70 years) for 12 weeks. The primary endpoint was the incidence of all bleeding events (major, clinically relevant non-major, and minor bleeds). Patients underwent CT and/or MRI to assess asymptomatic intracranial hemorrhage (ICH). Secondary endpoints included thromboembolic events and pharmacodynamic indices. The mean incidence of all bleeding events for edoxaban 30, 45, and 60mg, and warfarin was 18.5%, 22.4%, 27.7%, and 20.0%, respectively. There were no statistically significant differences among the edoxaban groups and no significant differences from the warfarin group. There were no asymptomatic ICH events in any group. One episode of cerebral infarction was observed in the edoxaban 45-mg group. Subgroup analysis suggested low body weight (≤60kg) was associated with higher bleeding risk. Edoxaban 30, 45, and 60mg QD in patients with NVAF was associated with a numerical increase in all bleeding across the dose range, but this was not statistically significant, nor was any dose compared with warfarin.