Background: The relative risks and benefits of starting or withholding antiep ileptic drug treatment in patients with few or infrequent seizures are unclear. We sought to compare policies of immediate versus deferred treatment in such pat ients and to assess the effects of these policies on short- term recurrence and long- term outcomes. Methods: We undertook an unmasked, multicentre, randomise d study of immediate and deferred antiepileptic drug treatment in 1847 patients with single seizures and early epilepsy. Outcomes comprised time to first, secon d, and fifth seizures; time to 2- year remission; no seizures between years 1 a nd 3 and between years 3 and 5 after randomisation; and quality of life. Analysi s was by intention to treat. Findings: 404 patients invited to join the trial di d not consent to randomisation; 722 were subsequently assigned immediate treatme nt with antiepileptic drugs and 721 were assigned deferred treatment. Immediate treatment increased time to first seizure (hazard ratio 1.4 [95% CI 1.2 to 1.7 ]), second seizure (1.3 [1.1 to 1.6]), and first tonic- clonic seizure (1.5 [1. 2 to 1.8]). It also reduced the time to achieve 2- year remission of seizures ( p=0.023). At 5- years follow- up, 76% of patients in the immediate treatment group and 77% of those in the deferred treatment group were seizure free betw een 3 and 5 years after randomisation (difference - 0.2% [95% CI - 5.8% to 5.5% ]). The two policies did not differ with respect to quality of life out comes or serious complications. Interpretation: Immediate antiepileptic drug tre atment reduces the occurrence of seizures in the next 1- 2 years, but does not affect long- term remission in individuals with single or infrequent seizures. Background: The relative risks and benefits of starting or withholding antiep ileptic drug treatment in patients with few or infrequent seizures are unclear. We sought to compare policies of immediate versus deferred treatment in such patients and to assess the effects of these policies on short- Methods: We undertook an unmasked, multicentre, randomise d study of immediate and deferred antiepileptic drug treatment in 1847 patients with single seizures and early epilepsy. Outcomes comprised of time to first, secon d, and fifth seizures; time to 2-year remission; no seizures between years 1 a nd 3 and between years 3 and 5 after randomisation; and quality of life. Analysi s was by intention to treat. Findings: 404 patients invited to join the trial di d not consent to randomisation; 722 were subsequently assigned immediate treatme nt with antiepileptic drugs and 721 were assigned deferred treatment. Immediate treatment increased time to first seiz ure (hazard ratio 1.4 [95% CI 1.2 to 1.7]), second seizure (1.3 [1.1 to 1.6]), and first tonic-clonic seizure At 5- years follow-up, 76% of patients in the immediate treatment group and 77% of those in the deferred treatment group were seizure free betwenen 3 and 5 years after randomisation (difference - 0.2% [95% CI - 5.8% to 5.5%]). The two policies did not differ with respect to quality of life out comes or serious complications. Interpretation: Immediate antiepileptic drug tre atment reduces the occurrence of seizures in the next 1- 2 years, but does not affect long- term remission in individuals with single or infrequent seizures.