Multiple sclerosis can follow very different patterns of evolution and variable rates of disability accumulation. This raises the issue whether it represents one or several distinct diseases. We assessed demographic and clinical characteristics in 1844 patients with multiple sclerosis that we categorized according to the classification of Lublin and Reingold (1996) into 1066 (58%) relapsing-remitting, 496 (27%) secondary progressive, 109 (6%) progressive relapsing and 173 (9%) primary progressive cases of multiple sclerosis. Relapsing-remitting and secondary progressive cases shared similar age at disease onset (median = 28.7 versus 29.5 years; P = 0.21), initial symptoms of the relapsing-remitting phase, degree of recovery from the first neurological episode, and time from the first to the second episode. By contrast, disease duration was twice as long in secondary progressive than in relapsing-remitting cases (mean +/- SD = 17.6 +/- 9.6 versus 8.7 +/- 8.6 years; P < 0.001). Progressive relapsing and primary progressive cases were essentially similar in their clinical characteristics. In patients experiencing a progressive course, median age at onset of progressive phase was similar in secondary progressive cases and in cases who were progressive from onset (39.1 versus 40.1 years; P = 0.47). The proportion of cases with superimposed relapses during progression was approximately 40% in both categories. Finally, the 1562 patients with an exacerbating-remitting initial course and the 282 patients with a progressive initial course of the disease were essentially similar with respect to the time course of disability accumulation from assignment to a given disability score, and the age at assignment of disability landmarks. These observational data suggest that the clinical phenotype and course of multiple sclerosis are age dependent. Relapsing-remitting disease can be regarded as multiple sclerosis in which insufficient time has elapsed for the conversion to secondary progression; secondary progressive forms as relapsing-remitting multiple sclerosis that has 'grown older'; and progressive from onset cases as multiple sclerosis 'amputated' from the usual preceding relapsing-remitting phase. Times to reach disability milestones, and ages at which these landmarks are reached, follow a predefined schedule not obviously influenced by relapses, whenever they may occur, or by the initial course of the disease, whatever its phenotype. This leads to a unifying concept of the disease in which primary and secondary progression might be regarded as essentially similar. From the clinical and statistical positions, multiple sclerosis might be considered as one disease with different clinical phenotypes rather than an entity encompassing several distinct diseases-the position of complexity rather than true heterogeneity.