Abstract Aims Persistence with direct oral anticoagulants (DOACs) has become a concern in non-valvular atrial fibrillation (NVAF) patients, but whether this affects prognosis is rarely studied. We investigated the persistence with oral anticoagulants (OACs) and its association with prognosis among a nationwide cohort of NVAF patients. Methods and results DOAC-naive NVAF patients who started to use DOACs for ischaemic stroke prevention between 2013 and 2018 were included using Dutch national statistics. Persistence with OACs was determined based on the presence of a 100-day gap between the last prescription and the end of study period. In 93 048 patients, 75.7% had a baseline CHA2DS2-VASc score of ≥2. The cumulative incidence of persistence with OACs was 88.1% [95% confidence interval (CI) 87.9–88.3%], 82.6% (95% CI 82.3–82.9%), 77.7% (95% CI 77.3–78.1%), and 72.0% (95% CI 71.5–72.5%) at 1, 2, 3, and 4 years after receiving DOACs, respectively. Baseline characteristics associated with better persistence with OACs included female sex, age range 65–74 years, permanent atrial fibrillation, previous exposure to vitamin K antagonists, stroke history (including transient ischaemic attack), and a CHA2DS2-VASc score ≥2. Non-persistence with OACs was associated with an increased risk of the composite outcome of ischaemic stroke and ischaemic stroke-related death [adjusted hazard ratio (aHR) 1.79, 95% CI 1.49–2.15] and ischaemic stroke (aHR 1.58, 95% CI 1.29–1.93) compared with being persistent with OACs. Conclusion At least a quarter of NVAF patients were non-persistent with OACs within 4 years, which was associated with poor efficacy of ischaemic stroke prevention. The identified baseline characteristics may help identify patients at risk of non-persistence.