OBJECTIVE: Little is known about how symptom severity in the various neurologic domains commonly affected by multiple sclerosis (MS) varies by age, sex, and race/ethnicity. METHODS: This was a retrospective study of patients with MS attending 2 tertiary centers in the New York City metropolitan area, who self-identified as White, African American (AA), or Hispanic American (HA). Disability was rated with Patient-Determined Disability Steps (PDDS) and symptom severity, with SymptoMScreen (SyMS), a validated battery for assessing symptoms in 12 domains. Analyses comparing race, sex, and age groups were performed using analysis of variance models and Tukey honestly significant difference tests to control the overall type I error. A multivariable model was constructed to predict good self-rated health (SRH) that included demographic variables, PDDS, and SyMS domain scores. RESULTS: The sample consisted of 2,622 patients with MS (age 46.4 years; 73.6% female; 66.4% White, 21.7% AA, and 11.9% HA). Men had higher adjusted PDDS than women (p = 0.012), but similar total SyMS scores. Women reported higher fatigue and anxiety scores, whereas men had higher walking and dexterity scores. AAs and HAs had higher symptom domain scores than Whites in each of the 12 domains and worse SRH. In a multivariable logistic model, only pain, walking, depression, fatigue, and global disability (PDDS), but not sex or race/ethnicity, predicted good SRH. CONCLUSIONS: AA and HA race/ethnicity was associated with higher overall disability, higher symptom severity in each of the 12 domains commonly affected by MS, and worse SRH relative to Whites. However, only symptom severity and disability, and not demographic variables, predicted good SRH.