Background: Approximately two-thirds of patients with newly diagnosed epilepsy become seizure-free after antiepileptic drug (AED) treatment. A crucial issue for these patients and their families, especially after a long period of seizure freedom, is when to stop their medications. Objective: The aim of this study was to identify the optimal timing of AED withdrawal in adults with focal epilepsy who had been seizure-free for ≥ 2 years. Methods: Adults with focal epilepsy who had been seizure-free for ≥ 2 years were recruited. Based on their decision to discontinue (withdrawal) or continue (non-withdrawal) AED treatment, patients were assigned to withdrawal or non-withdrawal subgroups according to the length of remission (2 to < 3 years, 3 to < 4 years, 4 to < 5 years and ≥ 5 years). The relapse risks of the withdrawal and corresponding non-withdrawal subgroups were compared, and the relative relapse risks were assessed in a Cox proportional hazard regression model. Results: A total of 213 eligible patients began to withdraw from AED treatment; 70 had been seizure-free for 2 to < 3 years, 62 had been seizure-free for 3 to < 4 years, 37 had been seizure-free for 4 to < 5 years and 44 had been seizure-free for ≥ 5 years. The figures for the corresponding non-withdrawal subgroups were 463, 334, 251 and 182, respectively. There was a significantly higher risk of seizure relapse in patients withdrawing from AEDs after 2 to < 5 years of seizure freedom than in the corresponding non-withdrawal controls, and the relative relapse risk was 3.052 (95% confidence interval [CI] 2.126–4.381; p < 0.001) for the seizure-free period of 2 to < 3 years, 3.617 (95% CI 2.384–5.488; p < 0.001) for 3 to < 4 years and 2.644 (95% CI 1.456–4.799; p = 0.001) for 4 to < 5 years. However, for patients who were seizure-free for ≥ 5 years, AED withdrawal did not significantly increase the risk of seizure relapse compared with that of patients continuing treatment (hazard ratio [HR] 1.362, 95% CI 0.634–2.926, p = 0.428). Compared with a seizure-free period of 2 to < 3 years, the relative relapse risk after AED withdrawal was significantly reduced only after being seizure-free for ≥ 5 years (HR 0.441, 95% CI 0.233–0.834; p = 0.012). Conclusion: Overall, for adults with focal epilepsy, withdrawal from AEDs significantly increased the risk of seizure relapse after being seizure-free for 2 to < 5 years, but might not increase the risk if the seizure-free period was ≥ 5 years.