<p>Antiseizure medication (ASM) withdrawal has long been a tricky problem haunting patients with epilepsy and their physicians. It is important to have an accurate evaluation of recurrence risk for epilepsy patients who are seizure-free and wish to discontinue ASMs by considering relevant predictors. The purpose of this review is to summarize potential predictors based on recent research in this field. The predictors of relapse can be classified into four categories: test results, epilepsy characteristics, treatment-related factors and patient characteristics. The principal predictors are age at epilepsy onset, types of seizures, epilepsy and epilepsy syndromes, duration of seizure-free period, the number of ASMs and abnormal electroencephalography (EEG).The decisions of ASM withdrawal should be based on predictors and patients’ conditions and expectations. A discussion of ASM withdrawal must be made among patients, caregivers and physicians to make sure the benefits outweigh the risks. Future studies should explore the underlying mechanisms of relapse and the interactions among different predictors. They should also aim to identify more risk factors, improve the quality of studies and include diverse outcome measurements.</p>