Direct oral anticoagulants (DOACs) have a favorable bleeding risk profile in patients with atrial fibrillation (AF). However, the safety of individual DOACs relative to warfarin for specific bleeding outcomes is less certain. We identified 423,450 patients with AF between 2013 to 2015 in the NCDR PINNACLE national ambulatory registry matched to the Centers for Medicare and Medicaid Services database. Outcomes included time to first major bleed, intracranial hemorrhage (ICH), major gastrointestinal bleed (GIB), or other major bleed. We estimated the association of OAC with bleeding using Cox proportional hazard models. The median duration of follow-up was 1.4 years. OACs were used in 64% of AF patients (66% warfarin, 15% rivaroxaban, 12% dabigatran, and 7% apixaban). A major bleeding event occurred in 6.9% of patients. Compared with warfarin users, fewer patients experienced ICH with the use of rivaroxaban (HR 0.73; 95% CI 0.64 to 0.84), dabigatran (HR 0.56; 95% CI 0.48 to 0.65), and apixaban (HR 0.70; 95% CI 0.55 to 0.90). The risk of major GIB was higher in rivaroxaban users (HR 1.20; 95% CI 1.12 to 1.27), and lower in dabigatran (HR 0.88; 95% CI 0.82 to 0.95) and apixaban (HR 0.84; 95% CI 0.74 to 0.95) users. For any DOAC versus warfarin, age (≥75 or <75 years) interacted with major bleeding (HR 0.93 vs 0.78; p <0.001), GIB (HR 1.10 vs 0.82; p <0.001), and other major bleeding (HR 0.93 vs 0.80; p <0.001). In conclusion, our results suggest that the safety of DOACs is superior to warfarin in AF patients, except with rivaroxaban and GIB. Age ≥75 years attenuated the relative safety benefits of DOACs.