AIMS: The aim of this article was to characterise the mechanisms underlying "breakthrough" ischaemic stroke (IS) or transient ischaemic attack (TIA) despite oral anticoagulation in patients with atrial fibrillation (AF). METHODS: We conducted a cross-sectional analysis of adults with non-valvular AF who experienced IS/TIA in the fifth Auckland Regional Community Stroke Study (ARCOS V: September 2020 to August 2021). Using clinical records, we collected data on demographics, comorbidities and peri-event anticoagulant dosing and intake. Anticoagulant adherence in the months preceding IS/TIA was categorised as good control if proportion of days covered (PDC) ≥80% for direct oral anticoagulant users or time in therapeutic range (TTR) ≥70% for warfarin users. IS/TIA mechanism was adjudicated using standardised criteria and classified as cardioembolic or non-cardioembolic, and patient characteristics were compared. RESULTS: Among 179 patients (76/179 [43%] female), 138/179 events (77%) were adjudicated as cardioembolic, while the remainder were attributed to competing mechanisms. Compared with non-cardioembolic events, cardioembolic aetiology was associated with younger median age (72 vs 81 years), lower proportions with good anticoagulant control (87/138 [63%] vs 34/41 [83%], p=0.017) and higher rates of peri-event missed or under-dosing (90/138 [65%] vs 18/41 [44%], p=0.014). In multivariable analysis, good control was independently associated with higher odds of a non-cardioembolic mechanism (adjusted odds ratio 3.67 [95% confidence interval 1.35-9.99], p=0.011). CONCLUSION: Most breakthrough IS/TIA events on oral anticoagulation were either associated with anticoagulant under-exposure (in patients with a cardioembolic aetiology) or a competing mechanism rather than anticoagulant failure. These findings highlight the importance of careful assessment to inform appropriate secondary prevention strategies.