We compared persistence of antiepileptic drugs (AEDs) including carbamazepine, oxcarbazepine, gabapentin, lamotrigine, topiramate, valproic acid, and phenytoin in an Asian population with epilepsy.A retrospective cohort study was conducted by analyzing Taiwan's National Health Insurance Research Database (NHIRD). Adult epilepsy patients newly prescribed with AEDs between 2005 and 2009 were included. The primary outcome was persistence, defined as the treatment duration from the date of AED initiation to the date of AED discontinuation, switching, hospitalization due to seizure or disenrollment from databases, whichever came first. Cox proportional hazard models were used to estimate the risk of non-persistence with AEDs.Among the 13,061 new users of AED monotherapy (mean age: 58 years; 60% men), the persistence ranged from 218.8 (gabapentin) to 275.9 (oxcarbazepine) days in the first treatment year. The risks of non-persistence in patients receiving oxcarbazepine (adjusted hazard ratio [HR], 0.78; 95% CI, 0.74-0.83), valproic acid (0.88; 0.85-0.92), lamotrigine (0.72; 0.65-0.81), and topiramate (0.90; 0.82-0.98) were significantly lower than in the carbamazepine group. Compared with carbamazepine users, the non-persistence risk was higher in phenytoin users (1.10; 1.06-1.13), while gabapentin users (1.03; 0.98-1.09) had similar risk. For risk of hospitalization due to seizure and in comparison with carbamazepine users, oxcarbazepine (0.66; 0.58-0.74) and lamotrigine (0.46; 0.35-0.62) users had lower risk, while phenytoin (1.35; 1.26-1.44) users had higher risk. The results remained consistent throughout series of sensitivity and stratification analyses.The persistence varied among AEDs and was better for oxcarbazepine, valproic acid, lamotrigine, and topiramate, but worse for phenytoin when compared with carbamazepine.