Objectives: To assess the risks of stroke and cardiovascular mortality for Aboriginal and non-Aboriginal Australians with atrial fibrillation. Design: Retrospective data linkage cohort study. Setting, participants: All people aged 20–84 years hospitalised with atrial fibrillation in Western Australia during 2000–2012. Main outcome measures: Stroke incidence rates and mortality after hospitalisation for atrial fibrillation, and 10-year risks of stroke and of cardiovascular and all-cause mortality. Results: Among 55 482 index admissions with atrial fibrillation, 7.7% of 20–59-year-old patients and 1.3% of 60–84-year-old patients were Aboriginal Australians. A larger proportion of Aboriginal patients aged 20–59 years had CHA2DS2-VASc scores of 2 or more (59.8% v 21.8%). In 20–59-year-old Aboriginal patients, the incidence during follow-up (maximum, 10 years; median, 7.1 years) of stroke (incidence rate ratio [IRR], 3.2; 95% CI, 2.5–4.1) and fatal stroke (IRR, 5.7; 95% CI, 3.9–8.9) were markedly higher than for non-Aboriginal patients. Stroke incidence was higher for 60–84-year-old patients, but the difference between Aboriginal and non-Aboriginal patients was smaller (IRR, 1.6; 95% CI, 1.3–2.0). Cardiovascular mortality during follow-up was also higher for 20–59-year-old Aboriginal patients (IRR, 4.4; 95% CI, 4.3–5.9). The hazards of stroke (adjusted HR [aHR], 1.67; 95% CI, 1.22–2.28) and cardiovascular mortality (aHR, 1.47; 95% CI, 1.18–1.83) in younger Aboriginal patients remained significantly higher after multivariable adjustment; age/sex, principal diagnosis of atrial fibrillation, and CHA2DS2-VASc score were the most influential factors. Conclusion: Stroke risk and cardiovascular mortality are markedly higher for Aboriginal than non-Aboriginal patients with atrial fibrillation, particularly for patients under 60. Strategies for providing evidence-based therapies and cardiovascular prevention to Aboriginal people with atrial fibrillation must be improved.