Abstract Aims People with atrial fibrillation (AF) frequently live with frailty, which increases the risk of mortality and stroke. This study reports the association between oral anticoagulation (OAC) and outcomes for people with frailty, and whether there is overall net benefit from treatment in people with AF. Methods and results Retrospective open cohort electronic records study. Frailty was identified using the electronic frailty index. Primary care electronic health records of 89 996 adults with AF and CHA2DS2-Vasc score of ≥2 were linked with secondary care and mortality data in the Clinical Practice Research Database (CPRD) from 1 January 1998 to 30 November 2018. The primary outcome was a composite of death, stroke, systemic embolism, or major bleeding. Secondary outcomes were stroke, major bleeding, all-cause mortality, transient ischaemic attack, and falls. Of 89 996 participants, 71 256 (79.2%) were living with frailty. The prescription of OAC increased with degree of frailty. For patients not prescribed OAC, rates of the primary outcome increased alongside frailty category. Prescription of OAC was associated with a reduction in the primary outcome for each frailty category [adjusted hazard ratio, 95% confidence interval, no OAC as reference; fit: vitamin K antagonist (VKA) 0.69, 0.64–0.75, direct oral anticoagulant (DOAC) 0.42, 0.33–0.53; mild frailty: VKA 0.52, 0.50–0.54, DOAC 0.57, 0.52–0.63; moderate: VKA 0.54, 0.52–0.56, DOAC 0.57, 0.52–0.63; severe: VKA 0.48, 0.45–0.51, DOAC 0.58, 0.52–0.65], with cumulative incidence function effects greater for DOAC than VKA. Conclusion Frailty among people with AF is common. The OAC was associated with a reduction in the primary endpoint across all degrees of frailty.