Twenty-seven patients with chronic progressive multiple sclerosis were treated with high-dose intravenous cyclophosphamide induction on either an impatient or outpatient basis. Following induction, patients were randomized to alternate-month outpatient "maintenance" or "no maintenance" therapy. These groups, as well as 24 nonrandomized control patients, were compared with each other after 12, 18, and 24 months of follow-up. All groups were similar in age, sex, duration of disease, and degree of disability before treatment. Fifty-nine percent of all cyclophosphamide-treated patients were stable at 12 months compared with 17% of all patients in the nonrandomized control group at 12 months. A statistically significant difference persisted at 18 and 24 months. A trend favoring maintenance therapy when compared with no maintenance therapy was evident at 12, 18, and 24 months, but was not statistically significant. Inpatient vs outpatient induction therapy failed to influence treatment outcome. Toxic side effects of nausea and vomiting presented a serious obstacle to maintenance therapy as administered in this protocol.