Patients with atrial fibrillation frequently benefit from anticoagulation to prevent stroke and systemic embolism. For decades, warfarin was the primary oral anticoagulant option despite its narrow therapeutic index requiring monitoring and drug-drug interactions. Dabigatran's recent availability provides practical advantages including no monitoring and fewer interactions; however, it lacks a convenient reversal agent for bleeding events. Currently, it is unclear what factors have driven anticoagulant utilization since dabigatran's introduction, and little real-world evidence on the agents' comparative effectiveness and safety is available. The objectives were to describe dabigatran and warfarin's utilization and switching patterns and assess their comparative effectiveness and safety. A cohort of non-valvular atrial fibrillation patients initiating anticoagulation from a large US database of commercial and Medicare supplement claims from 2009-2012 was extracted. We first examined factors associated with anticoagulant selection using a retrospective cohort design and multivariable regression. We then evaluated the effectiveness and safety of dabigatran compared with warfarin using multivariable Cox proportional hazards regression and propensity score weighting. Finally, we evaluated the clinical effects of switching anticoagulants compared with non-switching using a time-varying exposure design and multivariable Cox proportional hazards regression. Of the 64,935 patients included in the cohort, 32.5% used dabigatran. Dabigatran users were less likely to have high ischemic stroke or bleeding risk or other clinical comorbidities. Switching anticoagulation was also less frequent among patients with higher ischemic stroke or bleeding risk. Dabigatran was associated with a lower risk of ischemic stroke or venous thromboembolism, and no relation was seen between anticoagulant and harmful outcomes including bleeding events or acute myocardial infarction. However, dabigatran was also associated with a higher risk of gastrointestinal bleeding. Compared with non-switchers, no relation was seen between switching anticoagulants and an increased risk of stroke, systemic embolism, bleeding events, or myocardial infarction. Despite the rapid uptake of dabigatran, these results highlight that patients initiating dabigatran were generally healthier than those initiating warfarin. Dabigatran may be considered a safe and possibly more effective alternative to warfarin in patients with atrial fibrillation; despite encouraging results from the observed lack of increased adverse outcomes from switching anticoagulants, caution is still recommended.