Objectives. To examine the long-term risk of thromboembolism and bleeding in patients with atrial fibrillation comparing patients with and without recent breast cancer in subgroups with or without anticoagulation therapy, respectively. Design. Using nationwide registries, patients with breast cancer from 1998-2015 and subsequent atrial fibrillation within 3 years were stratified on anticoagulation and matched 1:3 on age, sex and comorbidities with atrial fibrillation patients without breast cancer. Risks of thromboembolism and bleeding were estimated by Aalen-Johansen and multivariable cox regression models. Results. Atrial fibrillation patients with and without anticoagulation were matched, respectively (201 and 525 with breast cancer matched with 603 and 1,575 without breast cancer). In patients with CHA2DS2-VASc-score >1 and anticoagulation the three years risks of thromboembolism were 4.2% (95% confidence interval (CI) 1.1-7.3) and 3.2% (CI 1.5-4.9) in patients with and without breast cancer. The risks of bleeding were 5.3% (CI 1.7-8.9) and 5.1% (CI 3.0-7.1), respectively. Breast cancer was associated with a similar risk of thromboembolism in patients with and without anticoagulation, respectively (Hazard ratio (HR) 1.10, CI 0.63-1.92 and HR 1.11, CI 0.82-1.50) and a similar risk of bleeding in patients with and without anticoagulation, respectively (HR 1.01, CI 0.56-1.84 and HR 0.85, CI 0.57-1.27) compared with the matched controls. Conclusions. Breast cancer was not associated with altered risk of thromboembolism or bleeding in patients with atrial fibrillation irrespective of treatment with anticoagulation. Our analyses suggest that atrial fibrillation diagnosed in patients with breast cancer should be considered as primary atrial fibrillation.