The ELDERCARE-AF trial showed that edoxaban at a very-low-dosage (VLD) of 15mg/day was more effective than a placebo at preventing stroke/systemic embolism without significantly increasing the risk of serious bleeding. We aimed to compare the effectiveness and safety for VLD non-vitamin K antagonist oral anticoagulants (NOACs) (edoxaban 15mg once daily, dabigatran 110 or 150 once daily, apixaban 2.5mg once daily, or rivaroxaban 10mg (without the diagnosis of chronic kidney disease) or less than 10mg once daily) versus regular-dosage (RD) NOACs (edoxaban 60/30mg once daily or other labeling-dosage NOACs) among a real-world cohort of elderly AF population similar to the ELDERCARE-AF cohort. In this nationwide retrospective cohort study from Taiwan National Health Insurance Research Database (NHIRD), we identified a total of 7294 and 4151 consecutive AF patients aged 80 years or older with a CHADS2 score≥2 who met the enrollment criteria (generally similar to ELDERCARE-AF) taking VLD and RD NOACs from June 1, 2012, to December 31, 2019, respectively. Propensity-score stabilized weighting (PSSW) was used to balance covariates across study groups. Patients were followed up from the first date of prescription for NOACs until the first occurrence of any study outcome, death, or until the end date of the study period (December 31, 2020). After PSSW, VLD NOAC was associated with a comparable risk of ischemic stroke/systemic embolism and major bleeding but a higher risk of major adverse limb events (MALEs) requiring lower limb revascularization or amputation (Hazard ratio [HR]:1.54, 95% confidential interval [CI]:1.09-2.18; p=0.014), venous thrombosis (HR:3.75,95%CI:1.56-8.97; p=0.003), and all-cause mortality (HR:1.21,95%CI:1.15-1.29; p<0.001) compared to RD NOACs. VLD NOACs showed worse outcomes in most net clinical benefits (NCOs). The main result was consistent based on on-treatment analysis or accounting for death as a competing risk. In general, the advantage of NCOs for the RD NOACs over VLD NOACs persisted in most high-risk subgroups, consistent with the main analysis (P for interaction>0.05). Use of VLD NOACs was associated with a greater risk of arterial and venous thrombosis, death as well as the composite outcomes, when compared to that of RD NOAC in high-risk elderly AF patients at increased bleeding risk. Thromboprophylaxis with RD NOAC is still preferable over VLD NOAC for the majority of elderly AF patients at increased bleeding risk.