Three characteristic features describe our understanding of the clinical course and outcome in multiple sclerosis: a validated statistical model of disease progression, wide interindividual variability, and a fixed rate of progression in an individual patient. However for and individual patient, it is still impossible to derive a precise and exact prediction of disease outcome, the only reliable method is the deterministic approach developed by Fog and Linneman in 1970 which consists in objective quantitative neurological examinations performed at three month intervals over several years in the same patient. In routine practice, this method in rather unrealistic. Bran MRI data with conventional techniques are also poorly discriminant. Conversely, it may be anticipated that new magnetic resonance techniques, more sensitive to axonal loss, demyelinization and gliosis, will provide reliable answers to this issue upon which therapeutic decisions depend.