The relapsing form of neuromyelitis optica (NMO) is characterized by recurrent optic neuritis and myelitis, usually leading to severe, permanent, relapse-related neurologic impairment (e.g., blindness, paraplegia) within 5 years. Aggressive therapy aimed at relapse prevention initiated soon after disease onset may be expected to have a relatively greater impact on early relapse-related disability in NMO than in typical MS. Early prediction of a relapsing course and subsequent disease severity would facilitate design and implementation of clinical trials of such therapies. A database of clinical and laboratory features of patients with NMO (n = 80) was used to develop potentially useful models predictive of a relapsing disease course and of subsequent disease severity as measured by survival. Predictors of a relapsing course were longer interattack interval between the first two clinical events (rate ratio [RR] = 2.16; per month increase), older age at onset (RR = 1.08; per year increase), female sex (RR = 10.0, female vs male), and less severe motor impairment with the sentinel myelitis event (RR = 0.48; per severity scale point increase). A history of other autoimmune disease (RR = 4.15; presence vs absence), higher attack frequency during the first 2 years of disease (RR = 1.21; per attack), and better motor recovery following the index myelitis event (RR = 1.84; per point increase) were associated with mortality due to relapsing NMO. These predictive models identify several clinical features, each available at diagnosis or early in the disease course, that predict relapsing disease and survival. These results may be useful to identify patients at high risk for severe, relapsing neuromyelitis optica in order to initiate early therapy for relapse prevention and to design clinical trials to study such interventions.