Optic neuritis (ON) is associated with numerous immune-mediated inflammatory diseases. After presenting for the first time with undifferentiated ON, up to 50% per cent of patients are subsequently diagnosed with multiple sclerosis (MS), sometimes many years later. Differentiating MS-ON from non-MS-ON in the acute setting is challenging but important. Initial treatment of MS-ON and non-MS ON differs, with the latter often requiring urgent immunosuppression. Diagnostic delay in non-MS ON impacts final visual prognosis. Over 200 common genetic variants are associated with MS. We sought to establish whether integrating an MS genetic risk score (MS-GRS) improves prediction of future MS in undifferentiated ON. Using data from the United Kingdom Biobank we showed that combining MS-GRS with demographic risk factors significantly improves MS prediction in patients with undifferentiated ON. In a combined risk model including sex, age and MS-GRS, one standard deviation of MS risk score increased the Hazard of MS 1.3-fold (95% confidence interval 1.09-1.59, P<0.005). Participants stratified into quartiles of predicted MS risk developed incident MS during cumulative follow-up at rates varying from 3.8% (95% CI 1.2–8.6%, lowest risk quartile) to 41.7% (95% CI 33.1–50.6%, highest risk quartile). This study indicates that using a combined model in clinical practice might enhance individual MS risk stratification, paving the way for precision-based ON treatment and earlier introduction of MS disease-modifying therapy.