Abstract Background Limited real world data show that rivaroxaban following dosage criteria from either ROCKET AF (15mg/day or 10mg/day if creatinine clearance (CrCl)<50ml/min) or J-ROCKET AF (15mg/day or 10mg/day if CrCl<50ml/min) are associated with comparable risks of thromboembolism and bleeding with each other in patients with non-valvular atrial fibrillation (NVAF). We aimed to study whether these observations differ between Asian and non-Asian subjects. Methods Databases were searched for adjusted observational study that compared relevant clinical outcomes of NVAF patients receiving off-label under-dosing rivaroxaban (10mg/day if CrCl>50ml/min), on-label dose rivaroxaban eligible for ROCKET AF or J-ROCKET AF, and off-label over-dosing rivaroxaban (20mg/day if CrCl<50ml/min). Results Eighteen studies were included. Rivaroxaban following J-ROCKET AF criteria (n=19,513) was associated with comparable risks of thromboembolism in the Asian subgroup (n=43,076), whereas rivaroxaban following J-ROCKET AF criteria (n=8555) was associated with higher risks of thromboembolism (hazard ratio (HR):1.27;95% confidential interval (CI):1.05–1.53) and all-cause mortality (HR:1.30;95%CI:1.05–1.60) compared with that of ROCKET AF criteria (n=23,139) in the non-Asian subgroup (n=31,694)(Figure). There was no differences in risks of major bleeding between rivaroxaban following J-ROCKET AF versus ROCKET AF criteria either in the Asian or non-Asian subgroup (Figure). Off-label underdosed rivaroxaban 10mg (n=6305) was associated with a higher risk of thromboembolism (HR:1.64;95%CI:1.28–2.11) but lower risk of major bleeding (HR:0.72;95%CI:0.57–0.90) compared with eligible dosage criteria (n=11,673). Off-label overdosed rivaroxaban 20mg (n=7572) was associated with worse clinical outcomes in the risks of thromboembolism (HR:1.42;95%CI:1.14–1.79), mortality (HR:1.41;95%CI:1.15–1.73) and major bleeding (HR:1.27; 95%CI:1.05–1.54; N=7 with I2=25%) compared with eligible dosage criteria (n=39,097). Conclusions Rivaroxaban dosing regimen following J-ROCKET criteria may serve as an alternative to ROCKET-AF criteria for the Asian population with NVAF, whereas the dosing regimen following ROCKET AF criteria was more favorable for the non-Asian population. Off-label underdosed rivaroxaban 10mg was associated with a higher risk of thromboembolism but a lower risk of major bleeding, while off-label overdosed rivaroxaban 20mg was associated with worse outcome in most clinical events.