It was the aim of this study to determine the efficacy and safety of vitamin K antagonists (VKAs) and acetylsalicylic acid (ASA) in patients with non-valvular atrial fibrillation (AF), with separate analyses according to predicted thromboembolic and bleeding risk. By individual level-linkage of nationwide registries, we identified all patients discharged with non-valvular AF in Denmark (n=132,372). For every patient, the risk of stroke and bleeding was calculated by CHADS2, CHA2DS2-VASc, and HAS-BLED. During follow-up, treatment with VKA and ASA was determined time-dependently. VKA consistently lowered the risk of thromboembolism compared to ASA and no treatment; the combination of VKA+ASA did not yield any additional benefit. In patients at high thromboembolic risk, hazard ratios (95% confidence interval) for thromboembolism were: 1.81 (1.73-1.90), 1.14 (1.06-1.23), and 1.86 (1.78-1.95) for ASA, VKA+ASA, and no treatment, respectively, compared to VKA. The risk of bleeding was increased with VKA, ASA, and VKA+ASA compared to no treatment, the hazard ratios were: 1.0 (VKA; reference), 0.93 (ASA; 0.89-0.97), 1.64 (VKA+ASA; 1.55-1.74), and 0.84 (no treatment; 0.81-0.88), respectively. There was a neutral or positive net clinical benefit (ischaemic stroke vs. intracranial haemorrhage) with VKA alone in patients with a CHADS2 score of = 0, and CHA2DS2-VASc score of = 1. This large cohort study confirms the efficacy of VKA and no effect of ASA treatment on the risk of stroke/thromboembolism. Also, the risk of bleeding was increased with both VKA and ASA treatment, but the net clinical benefit was clearly positive, in favour of VKA in patients with increased risk of stroke/thromboembolism.