Background: The Framingham Study from 1991 showed a clear correlation between atrial fibrillation (AF) and ischemic stroke, where patients with AF had an almost fivefold increase in risk of stroke compared with patients without AF. Since then, several trials have evaluated different antithrombotic treatments to reduce the risk of stroke in patients with AF. Other trials have investigated factors that increase the risk of stroke in patients with AF and risk score systems have been developed to categorize patients into low or increased risk of stroke to help clinicians to decide which patients benefit from antithrombotic treatment and in whom it can be abstained, not to expose patients with low stroke risk to an increased risk of bleeding conferred by antithrombotic treatment. The aims of this thesis were: [1] to evaluate if a warfarin dosing algorithm can increase hit rate and decrease mean error compared with manually changed doses; [2] to assess the prevalence and net clinical benefit of aspirin as monotherapy for stroke prevention in AF; [3] to investigate the risk of thromboembolic and haemorrhagic complications within 30 days after electrical cardioversion (ECV) of AF in patients with and without oral anticoagulation (OAC) pre-treatment; and [4] to assess the proportion of patients discontinuing OAC after pulmonary vein isolation (PVI), identify factors predicting stroke after PVI and to investigate risk of complications after PVI with and without OAC. Materials and methods: All studies are retrospective and based on data from Swedish national quality registries. In paper I, data from Auricula was used to compare the resulting INR values after algorithmic warfarin dose suggestions and manually changed doses. In paper II data was extracted from the Swedish National Patient Register, the Dispensed Drugs Register and the Cause of Death Register. Patients with aspirin treatment were compared with patients without any antithrombotic treatment regarding risk of thromboembolic and haemorrhagic complications. In paper III data was collected from the Swedish National Patient Register and the Dispensed Drugs Register to examine risk of complications (thromboembolic and haemorrhagic events) within 30 days after cardioversion, comparing patients with and without oral anticoagulation pre-treatment. In paper IV data from six different Swedish national quality registries were used (Swedish Catheter Ablation Register, Auricula, Swedish National Patient Register, Dispensed Drugs Register, Cause of Death Register and Riksstroke). Patients undergoing pulmonary vein isolation (PVI) were investigated for adherence to guidelines regarding oral anticoagulation, predictors for stroke after PVI, as well as risk of ischemic stroke or intracranial haemorrhage after PVI in patients with and without treatment. Results: Paper I showed that a computerized dosing algorithm for warfarin in most cases perform as well or better compared with doses that have been changed manually, with a better hit-rate (0.72 vs. 0.67) and a lower mean error (0.44 vs. 0.48). Paper II showed that 32% of 182.678 patients with a diagnosis of AF were on monotherapy with aspirin for stroke prevention. A total of 115.185 patients were included, 58.671 with aspirin treatment and 56.514 without antithrombotic treatment at baseline. After stratification after CHA2DS2-VASc score and after multivariable adjustment, aspirin treatment did not confer a decrease in thromboembolic events. After propensity score mathcing, rate of ischemic stroke was 7.4%/year (95% CI 7.1-7.6) in aspirin treated patients and 6.6%/year (95% CI 6.4-6.9) in patients without antithrombotic treatment. In paper III 22.874 patients undergoing electrical cardioversion were included, 10.722 with and 12.152 without OAC pre-treatment. In patients with low stroke risk (CHA2DS2-VASc 0-1), no thromboembolic complication was seen within 30 days after cardioversion. In patients with CHA2DS2-VASc ≥2, the risk of thromboembolic complications was increased when no oral anticoagulation pre-treatment was used, results that remained after propensity score matching. No difference regarding haemorrhagic complications was seen. Paper IV included a total of 1585 patients undergoing PVI with a mean follow up of 2.6 years. Adherence to current guidelines regarding oral anticoagulation was good in patients with CHA2DS2-VASc ≥2. Previous ischemic stroke was a predictor for a new stroke after PVI. In patients with CHA2DS2-VASc ≥2 stroke risk was increased in patients discontinuing OAC compared to those continuing OAC (1,60%/year vs. 0.34%/year). Conclusion: Oral anticoagulation is still underutilized for prevention of stroke and systemic embolism in patients with atrial fibrillation. Patients with risk factors for stroke (CHA2DS2-VASc ≥2p) benefit from continuous oral anticoagulation treatment to prevent stroke, also in conjunction with electrical cardioversion and after pulmonary vein isolation. If warfarin is chosen, a computerised dosing algorithm can facilitate and standardize warfarin dosing and lead to better resulting INR values than manually changed doses. Aspirin should not be used for stroke prevention in patients with atrial fibrillation.