To find out whether non-specific immunosuppression is beneficial in multiple sclerosis (MS) a randomised, placebo-controlled, single-masked trial was carried out in nine university centres. 168 patients with clinically or laboratory-supported definite MS in progressive phase (deterioration by at least 1.0 on the expanded disability status scale [EDSS] in the previous year) were randomised to receive intravenous cyclophosphamide and oral prednisone (n = 55); daily oral cyclophosphamide, alternate day prednisone (22 weeks), and weekly plasma exchange (20 weeks) (n = 57); or placebo medications and sham plasma exchange (n = 56). All patients were followed for at least 12 months (mean 30.4 months) by a monitoring neurologist, who was aware of treatment allocation, and an evaluating neurologist, who was not. The primary analysis was a comparison of rates of treatment failure (worsening of evaluating neurologist's assessment of EDSS by 1.0 or more on two consecutive 6-monthly assessments). There were no significant differences among the groups in this primary analysis (19 [35%] treatment failures with cyclophosphamide; 18 [32%] with plasma exchange; 16 [29%] with placebo). Nor were there any differences in the proportions improved, stabilised, or worsened at each 6 month assessment or in the mean change in the EDSS at the final assessment (0.81 cyclophosphamide; 0.69 plasma exchange; 0.69 placebo). A slight trend favouring the plasma exchange group at 12-24 months of follow-up was not sustained at the final assessment. This study fails to confirm previous reports that immunosuppressive treatments result in stabilisation or improvement in progressive MS.