BACKGROUND: Acute demyelinating optic neuritis, a common feature of multiple sclerosis, can damage vision through neurodegeneration in the optic nerve and in its fibres in the retina. Inhibition of voltage-gated sodium channels is neuroprotective in preclinical models. In this study we aimed to establish whether sodium-channel inhibition with phenytoin is neuroprotective in patient with acute optic neuritis. METHODS: We did a randomised, placebo-controlled, double-blind phase 2 trial at two UK academic hospitals in London and Sheffield. Patients with acute optic neuritis aged 18-60 years, presenting within 2 weeks of onset, with visual acuity of 6/9 or worse, were randomly assigned (1:1) by minimisation via a web-based service to oral phenytoin (maintenance dose 4 mg/kg per day if randomised before or on July 16, 2013, and 6 mg/kg per day if randomised on or after July 17, 2013) or placebo for 3 months, stratified by time from onset, centre, previous multiple sclerosis diagnosis, use of disease-modifying treatment, and use of corticosteroids for acute optic neuritis. Participants and treating and assessing physicians were masked to group assignment. The primary outcome was retinal nerve fibre layer (RNFL) thickness in the affected eye at 6 months, adjusted for fellow-eye RNFL thickness at baseline, analysed in a modified intention-to-treat population of all randomised participants who were followed up at 6 months. Safety was analysed in the entire population, including those who were lost to follow-up. The trial is registered with ClinicalTrials.gov, number NCT 01451593. FINDINGS: We recruited 86 participants between Feb 3, 2012, and May 22, 2014 (42 assigned to phenytoin and 44 to placebo). 29 were assigned to phenytoin 4 mg/kg and 13 to phenytoin 6 mg/kg. Five participants were lost to follow-up, so the primary analysis included 81 participants (39 assigned to phenytoin and 42 to placebo). Mean 6-month RNFL thickness in the affected eye at 6 months was 81.46 μm (SD 16.27) in the phenytoin group (a mean decrease of 16.69 μm [SD 13.73] from baseline) versus 74.29 μm (15.14) in the placebo group (a mean decrease of 23.79 μm [13.97] since baseline; adjusted 6-month difference of 7.15 μm [95% CI 1.08-13.22]; p=0.021), corresponding to a 30% reduction in the extent of RNFL loss with phenytoin compared with placebo. Treatment was well tolerated, with five (12%) of 42 patients having a serious adverse event in the phenytoin group (only one, severe rash, was attributable to phenytoin) compared with two (5%) of 44 in the placebo group. INTERPRETATION: These findings support the concept of neuroprotection with phenytoin in patients with acute optic neuritis at concentrations at which it blocks voltage-gated sodium channels selectively. Further investigation in larger clinical trials in optic neuritis and in relapsing multiple sclerosis is warranted. FUNDING: US National Multiple Sclerosis Society, Multiple Sclerosis Society of Great Britain and Northern Ireland, Novartis, UK National Institute for Health Research (NIHR), and NIHR UCLH/UCL Biomedical Research Centre.