The clinical data of 309 patients with definite multiple sclerosis were recorded in the European data base for multiple sclerosis (EDMUS) to determine the prognostic significance of several demographic and clinical variables. An interview with closed questions structured according to standardised criteria of disease phases and courses was used to assess the clinical course. The reliability was evaluated by four trained neurologists in a sample of 33 patients with multiple sclerosis. Both the within and between rater agreement on data collection was fair to high for the historical variables (K = 0.33-1). Between rater agreement was more variable for the evaluation of 12 different EDMUS event categories (K = 0.3-0.95). The predictive model for the time to reach a secondary progression showed that an age at onset older than 25 (p = 0.006) and an event at onset followed by disability > or = 3 on the Kurtzke expanded disability status scale (EDSS; p = 0.004) were the most unfavourable clinical variables in 249 patients with relapsing remitting (180) or relapsing progressive (69) courses. In the 69 patients with relapsing progressive disease, the time to reach severe disability (EDSS > or = 6) was negatively influenced by a first interval between attacks shorter than one year, a number of bouts with EDSS > 2 in the first two years of the disease, and involvement of the pyramidal system at onset (p < 0.05). In 60 patients with chronic progressive disease this outcome was negatively influenced by pyramidal, brainstem, and sensory involvement at onset (p < 0.01).