There are no generally effective disease-modifying drugs for progressive forms of multiple sclerosis (MS). Some MS centres use cyclophosphamide (CYC) in secondary progressive (SP) forms of MS, especially after interferon beta-1b (INFbeta-1b) treatment failure. Moreover, there are currently no approved drugs for primary progressive (PP) MS. Using the collected data of patients with progressive MS, we studied clinical patterns that predicted a good response to CYC treatment. Secondly, we compared the therapeutic response of SPMS and PPMS patients to the treatment. Data from 490 MS patients were collected. All patients presented an SP (n = 362) or PP (n = 128) form of the disease and 476 had been treated for at least one year with a monthly pulse of CYC associated with methylprednisolone (MP). CYC treatment was justified because of at least a 1-point worsening on the Expanded Disability Status Scale (EDSS) during the previous year. The EDSS score was assessed at baseline and after 6 months (M6) and 12 months (M12) of treatment. After 12 months of CYC treatment, 78.6% of SPMS and 73.5% of PPMS patients had stabilised or had an improved EDSS score. Response to CYC was not significantly different in the two progressive forms of MS. Twenty-two patients presented noticeable drug side effects, one of whom withdrew from the treatment due to intolerance. Patients with an improved EDSS at M12 had a shorter mean progressive time course (5.1 years) than patients who stabilised or worsened (7.1 years) (p = 0.02). We also observed that poor responders at M6 were also poor responders at M12 (p < 0.001). This large cohort study showed that CYC treatment was well tolerated and suggested that a better response occurred in cases with a short progressive time course. We did not find any difference in treatment response between the two progressive forms of MS. To date, no treatment is approved for PPMS and we therefore propose a trial to test the use of CYC treatment early in the course of the disease in PPMS patients with disability progression.