Limited real-world data are available regarding the comparative safety of non-vitamin K antagonist oral anticoagulants (NOACs). The objective of this retrospective claims observational cohort study was to compare the risk of bleeding among non-valvular atrial fibrillation (NVAF) patients prescribed apixaban, dabigatran, or rivaroxaban. NVAF patients aged ≥18 years with a 1-year baseline period were included if they were new initiators of NOACs or switched from warfarin to a NOAC. Cox proportional hazards modelling was used to estimate the adjusted hazard ratios of any bleeding, clinically relevant non-major (CRNM) bleeding, and major inpatient bleeding within 6 months of treatment initiation for rivaroxaban and dabigatran compared to apixaban. Among 60,227 eligible patients, 8,785 were prescribed apixaban, 20,963 dabigatran, and 30,529 rivaroxaban. Compared to dabigatran or rivaroxaban patients, apixaban patients were more likely to have greater proportions of baseline comorbidities and higher CHA2DS2-VASc and HAS-BLED scores. After adjusting for baseline clinical and demographic characteristics, patients prescribed rivaroxaban were more likely to experience any bleeding (HR: 1.35, 95% confidence interval [CI]: 1.26–1.45), CRNM bleeding (HR: 1.38, 95% CI: 1.27–1.49), and major inpatient bleeding (HR: 1.43, 95% CI: 1.17–1.74), compared to patients prescribed apixaban. Dabigatran patients had similar bleeding risks as apixaban patients. In conclusion, NVAF patients treated with rivaroxaban appeared to have an increased risk of any bleeding, CRNM bleeding, and major inpatient bleeding, compared to apixaban patients. There was no significant difference in any bleeding, CRNM bleeding, or inpatient major bleeding risks between patients treated with dabigatran and apixaban.