The study's objectives were to assess the predictive significance of different sets of demographic, clinical and extraclinical variables in identifying multiple sclerosis patients with various risk levels of worsening during the follow-up, in order to provide clues to inclusion criteria and selection of primary clinical end-points in therapeutic trials. Two hundred and twenty-four patients at their first diagnosis of multiple sclerosis admitted to our Department between 1983 and 1990 were prospectively followed-up until the end of 1996. We considered as end-points time to reach non-reversible disability levels corresponding to EDSS scores of 4.0 and 6.0 and the beginning of a secondary progressive phase in the relapsing-remitting subgroup of patients. For the statistical treatment of our data we used the Kaplan-Meier survival curves and the Cox regression analysis. An initially progressive course and higher basal EDSS scores proved to be the best predictors of unfavorable prognosis; a greater number of functional systems involved at onset as well as higher residual deficits in pyramidal, visual, sphincteric and cerebellar systems were other factors predictive of a poor outcome, whereas sensory system involvement turned out to be favorable. In the relapsing-remitting subgroup, a longer first inter-attack interval was associated with a better prognosis; however, overall number of relapses in the first two years of the disease was of no prognostic value. The presence of oligoclonal banding in the cerebrospinal fluid and a cerebral MRI 'strongly suggestive' or 'suggestive' of MS in the early phases of the disease were associated with a higher probability of a worse outcome.