Objective: To determine if non-myeloablative HSCT is superior to continuing DMD therapy. Background: We previously reported that non-myeloablative HSCT can be performed safely and is accompanied by long-term improvement in neurological disability in patients with RRMS. Design/Methods: Patients on stable DMDs with > 2 relapses within the prior 12 months were randomized (1:1) to treatment with cyclophosphamide and rabbit anti-thymocyte globulin followed by hematopoietic stem cell infusion or to a control arm with continued treatment with the most appropriate DMD as judged by their treating neurologist. The primary endpoint was treatment failure defined as an increase in EDSS, assessed by a blinded evaluating neurologist, by at least 1.0 point sustained for at least 6 months. Patients on DMDs who failed after at least 1 year of treatment were allowed to crossover to HSCT. Results: 110 patients were randomized, 55 to each arm. Three HSCT patients were withdrawn: two for failing enrollment criteria, one for recurrent infections occurring before transplant. Five control patients were withdrawn after soliciting transplants at other centers. No deaths occurred and no CTC grade 4 non-hematopoietic toxicities occurred in the transplant arm. DMDs (number of patients) used in the control arm were: natalizumab (22), dimethyl fumerate (18), fingolimod (13) interferons (10), glatiramer acetate (8), and mitoxantrone (5). Other immune drugs used in the control arm were corticosteroids (39), cyclophosphamide (2) and rituximab (2). With a mean follow up of 3 years (range 1 to 5 years), treatment failure was 60% (30 of 50) for control arm and 6% (3 of 52) for HSCT (P Conclusions: HSCT for RRMS with > 2 relapses a year was superior to continued DMDs. Disclosure: Dr. Burt has nothing to disclose. Dr. Balabanov has nothing to disclose. Dr. Snowden has nothing to disclose. Dr. Sharrack has nothing to disclose. Dr. Oliveira has nothing to disclose. Dr. Burman has nothing to disclose.