INTRODUCTION: There is limited information comparing the off-therapy relapse rates of patients discontinued tenofovir alafenamide (TAF) to those stopping entecavir or tenofovir disoproxil fumarate (TDF). METHODS: A total of 805 hepatitis B e antigen-negative patients without cirrhosis receiving entecavir (n = 406), TDF (n = 260), or TAF (n = 139) were enrolled. Propensity score matching method was applied to eliminate the significant differences in clinical characteristics. RESULTS: The cumulative incidences of virological relapse, clinical relapse, and retreatment at 96 weeks were higher in the off-TAF group (89.6%, 70.3%, and 59.2%, respectively) than that in the off-entecavir group (65.9%, 42.8%, and 28.8%, respectively) or the off-TDF group (73.7%, 49.8%, and 35.7%, respectively). The median time to clinical relapse was much earlier for off-TAF patients than for off-entecavir or off-TDF (median 14, 57, and 26 weeks, respectively), and these findings persisted even after propensity score matching. Multivariate analysis indicated that TAF therapy was an independent risk factor of virological relapse, clinical relapse, and retreatment when compared with entecavir or TDF. Hepatitis B surface antigen levels at end of treatment were predictive of virological, but not clinical, relapse in the off-TAF group, although this group had a lower rate of severe hepatitis on clinical relapse than the off-TDF group. Finally, there was no significant difference in the hepatic decompensation rate among the entecavir, TDF, and TAF groups. DISCUSSION: There is an earlier and higher hepatitis B virus relapse rate in patients who discontinue TAF therapy than in comparable patients discontinuing entecavir or TDF therapy. Close monitoring is necessary after TAF withdrawal, particularly in the first 3 months.