Three groups of 10 patients each with multiple sclerosis (MS) in a progressive phase were openly matched on the basis of age and invalidity (DSS Kurtzke). Variance analysis showed no significant difference between them for the main MS features. Group 1 received cyclophosphamide for 3 weeks (mean total dose: 152 mg/kg) with methylprednisolone (mean total dose: 2.77 g). Group 2 had a mean number of 9 plasma exchanges prior to a cyclophosphamide-methylprednisolone regimen similar to Group 1 (mean total dose of cyclophosphamide: 160 mg/kg and of methylprednisolone: 3.16 g). Group 3 was made up of controls. At three years, the proportion of stabilized and improved cases was 6/10 in group 1, 9/10 in Group 2 (statistically significant when compared with Group 1), and 0/10 in Group 3. The study of the variations of invalidity (DSS gains) showed a clear significant benefit in the treated groups when compared to controls, but no difference between the treated groups. Longitudinal studies showed that the mean therapeutic benefit was about 2.5 years. The role of cyclophosphamide and of plasma exchanges in these results is discussed.