Background The clinical course of multiple sclerosis (MS) evolves over many years. Its prognosis is highly variable among affected individuals, i.e. while some suffer from early severe disabilities, others remain ambulatory and functional for many years. We used Multiple Sclerosis Severity Score (MSSS) and the new classification for MS severity Herbert et al. introduced in 2006 according to MSSS, to investigate some clinical and demographic factors as potential indicators of disease severity in in MS. Methods During a six-month period, patients with definite MS according to the revised McDonald's criteria who referred to three neurology and MS clinics in Tehran (Iran) were included in the study. All patients were interviewed and examined by a neurology resident who had been trained for employing the Expanded Disability Status Scale (EDSS). For each patient, MSSS was determined by using EDSS and disease duration. Results Overall, 338 (266 female and 72 male) patients were enrolled. Among demographic features, gender, younger age at onset, positive family history, and parental consanguinity were not associated with disease severity. Education was weakly associated with disease severity. Among clinical factors, presenting symptoms such as poly-symptomatic attacks, walking difficulty, and upper and lower extremity dysfunction were associated with more disability while presentation with optic neuritis had better prognosis. Complete recovery after the first attack, longer interval between the first and second attacks, lower number of symptoms at presentation, shorter duration of attacks, and relapsing-remitting course were associated with less disability and better prognosis. These results were noticed in ordinal logistic regression. However when multiple logistic regression was performed, the strongest determinant of disease severity was disease course with odds ratio (OR) = 49.12 for secondary progressive course and OR = 53.25 for primary progressive (± relapse) course. Walking difficulty as the presenting symptom had a borderline association with disease severity (OR = 2.31; P = 0.055). Increased number of onset symptoms was associated (but not significantly) with more severe disease. Conclusion Early prediction of disease severity by demographic and clinical features is currently impossible. We need to determine stronger predictors, possibly a combination of demographic, clinical, biomarkers, and imaging findings.