Four direct oral anticoagulants (DOACs) are used in Japan (edoxaban, rivaroxaban, apixaban, and dabigatran); however, few studies have examined the long‐term treatment persistence of these DOACs. Furthermore, the factors associated with persistence remain unclear. This single‐center, retrospective cohort study enrolled participants who were newly prescribed the 4 DOACs between January 1, 2012, and April 30, 2020. We assessed the treatment persistence rate by calculating the cumulative incidence rate of prescription switch or discontinuation for 5 years from the initial prescription. The factors associated with persistence were examined using multivariate analysis. The edoxaban was used as a reference for comparison with the other DOACs. The persistence rate at 5 years was 52.9% for all DOACs, including 67.0%, 51.6%, 50.2%, and 37.0% for edoxaban, rivaroxaban, apixaban, and dabigatran, respectively. Multivariate analysis revealed that age >65 years (hazard ratio [HR], 0.62 [95%CI, 0.41–0.93]), chronic kidney disease (HR, 1.63 [95%CI, 1.11–2.39]), baseline hemoglobin (HR, 0.85 [95%CI, 0.78–0.93]), diabetes mellitus (HR, 0.51 [95%CI, 0.29–0.93]), and type of DOACs (rivaroxaban: HR, 1.81 [95%CI, 1.03–3.18]; apixaban: HR, 2.00 [95%CI, 1.15–3.48]; and dabigatran: HR, 2.84 [95%CI, 1.66–4.86]) were significantly associated with nonpersistence at 1 year. At 5 years, diabetes mellitus (HR, 0.60 [95%CI, 0.37–0.97]) and type of DOAC were significantly associated with nonpersistence (rivaroxaban: HR, 1.79 [95%CI, 1.09–2.94]; apixaban: HR, 2.04 [95%CI, 1.26–3.31]; and dabigatran: HR, 2.76 [1.73–4.42]). Long‐term treatment persistence differed according to the type of DOAC, with edoxaban exhibiting the highest level of persistence. The factors associated with persistence may change over the treatment course, but larger studies are required to generalize our findings.