OBJECTIVES The great majority of patients with atrial fibrillation (AF) require lifelong antithrombotic therapy for prevention of stroke. For optimal treatment, it is important to investigate treatment discontinuations, as those might be associated with increased hospitalization rates.  As potential indicator for unstable anticoagulation efforts we assessed hospitalization rates for non-vitamin K antagonist oral anticoagulants (NOAC) and for warfarin under real-life practice conditions. METHODS PREFER in AF (The PREvention oF thromboembolic events – European Registry in Atrial Fibrillation) is a prospective non-interventional disease registry of patients with AF in 7 countries in Western Europe. Discontinuation rates were assessed looking at all patients (n=6412), whereas hospitalization rates reported at Baseline (BL) and 1 year Follow-Up (FU) focused on two groups: patients treated with warfarin (BL=1379, FU=1571) and patients treated with NOAC (BL=194, FU=424). Descriptive statistics were applied. RESULTS Out of 6412 patients, 9 to 18% of patients treated with vitamin K antagonists (VKA) discontinued therapy (warfarin 10.3%, phenprocoumon 9%, acenocoumarol 10.4%, fluinidone 18%). The discontinuation rates for patients on NOAC were 9.4% for rivaroxaban and 10.3% for dabigatran, respectively. Reported hospitalization rates (irrespective of reason) were 19.8%/10.0% for NOAC, and 24.1%/11.6% for warfarin at BL or FU, respectively.  Mean duration of hospitalization was 1.3±4.0/ 0.9 ±5.1 days for NOAC, and 2.0±6.1/ 1.0 ±4.2 days for warfarin. CONCLUSIONS No major differences were observed in the discontinuation rates between VKA and NOAC on the class level nor the various available NOAC drugs. Hospitalization rates and corresponding number of days in hospital reported at BL were substantially higher compared to FU. Patients on NOAC had lower hospitalization rates and less days in hospital compared to those on warfarin. Further analyses are needed to explore the reasons for hospitalizations.