The effect of edoxaban on plasma prothrombin fragment 1+2 (PTF1+2), a sensitive maker of in vivo thrombin generation, has not been fully investigated in non-valvular atrial fibrillation (NVAF). We compared plasma PTF1+2 levels between 25 NVAF patients receiving warfarin and 100 NVAF patients receiving edoxaban and additionally analyzed the association between plasma PTF1+2 levels and the dose of edoxaban. Plasma PTF1+2 levels were significantly higher in patients receiving edoxaban than in those receiving warfarin (141.5 ± 50.0 pmol/l vs 93.1 ± 55.7 pmol/l, p <0.001). The prevalence of plasma PF1+2 levels above the upper limit (229 pmol/l) of the normal range did not differ between the 2 groups (4% vs 4%), whereas the prevalence of plasma PTF1+2 levels below the lower limit (69 pmol/L) of the normal range was significantly lower in patients receiving edoxaban than in those receiving warfarin (1% vs 48%, p <0.001). Multiple linear regression analysis identified age and warfarin treatment as independent variables associated with the plasma PTF1+2 level. In a subgroup analysis, plasma PTF1+2 levels were significantly higher in 58 receiving edoxaban of 30 mg/day than in 42 receiving edoxaban of 60 mg/day (157.6 ± 50.8 pmol/l vs 121.6 ± 39.8 pmol/l, p = 0.01); however, after adjusting for confounding factors, the dose of edoxaban was not independently associated with the plasma PTF1+2 level. In conclusion, edoxaban sufficiently inhibits thrombin generation unrelated to its dose in NVAF, although its inhibitory effect is weaker compared to warfarin.