SUMMARYObjective: We performed a retrospective, prognostic analysis of a cohort of patients with epilepsy according to time of relapse after four seizure-free years. Methods: Planned withdrawal of antiepileptic drugs (AEDs) and at least 3 years of follow-up after AED discontinuation were performed. The following two groups were assessed: (1) an early relapse (ER) group of patients who experienced recurrence during AED withdrawal and (2) a late relapse (LR) group of patients who experienced recurrence after completion of the AED discontinuation process. After dichotomization, the relapse rate, prognostic factors, and their impacts for each group were compared with those of a group of patients who continued to be seizure-free after AED withdrawal (SF group) using multiple logistic regression analysis. The AED intake mode was also analyzed. Results: Two hundred seventeen (64.6%) of the 336 total patients experienced relapse. One hundred thirty-nine patients (41.4%) and 78 patients (23.2%) were included in the LR and ER groups, respectively. Symptom duration >120 months showed the strongest negative prognostic impact as demonstrated by the 4.7-fold higher risk of recurrence in the ER group compared with the SF group. Additional factors with a negative prognostic impact included an age at epilepsy onset of ≤20 years and the presence of localization-related epilepsy. No reliable predictor between the SF and LR groups was revealed. After exclusion of the SF group, post hoc analysis according to age at epilepsy onset and symptom duration showed that the above-mentioned negative prognostic factors significantly affected the relapse patterns of the LR and ER groups. Significance: The results suggest that longer symptom duration, which could be associated with intrinsic reactivation of epilepsy, is the strongest negative prognostic factor for relapse. Relapse after AED withdrawal in prolonged follow-up of seizure-free patients is one aspect of the natural history of epilepsy. KEY WORDS: Early relapse, Late relapse, Symptom duration, Age at epilepsy onset, Natural history of epilepsy.Complete withdrawal of an AED (antiepileptic drug) without relapse is the final goal of epilepsy treatment. Understanding of when to submit seizure-free patients to AED withdrawal has long been an important issue. Approximately 70% of all newly diagnosed patients with epilepsy become seizure-free with the use of an AED, 1 and 67% of patients go into terminal remission.2 However, there have been broad differences among studies 3-6 regarding relapse rates following AED withdrawal, which might be due to confounding social factors associated with each patient. Risk-benefit assessments of AED withdrawal are highly