OBJECTIVE: Since 1996, autologous hemopoietic stem cell transplantation (HSCT) has been used to treat severe rheumatoid arthritis (RA). To date, published reports have been individual cases or series containing small numbers. This study combined the worldwide experience in a single analysis. METHODS: The Autoimmune Disease Databases of the European Group for Blood and Marrow Transplantation (EBMT) and the Autologous Blood and Marrow Transplant Registry (ABMTR) were used to identify patients with RA treated with autologous HSCT. Further information relating to patient and treatment-specific variables was obtained by questionnaire. RESULTS: Seventy-six patients were registered from 15 centers. Seventy-three patients had received autologous HSCT, and in 3 patients hematopoietic stem cells (HSC) were mobilized but not transplanted. Transplanted patients (median age 42 yrs, 74% female, 86% rheumatoid factor positive) had been previously treated with a mean of 5 (range 2-9) disease modifying antirheumatic drugs (DMARD). Significant functional impairment was present, with a median Health Assessment Questionnaire (HAQ) score of 1.4 (range 1.1-2.0) and Steinbrocker score mean 2.39 (SD 0.58). The high dose treatment regimen was cyclophosphamide (CYC) alone in the majority of patients, mostly 200 mg/kg (n = 62). Seven patients received anti-thymocyte globulin (ATG) in addition to CYC, 2 patients busulfan and CYC (BuCYC), and one patient CYC with total body irradiation and ATG. One patient received fludarabine with ATG. Following treatment, one patient received bone marrow but the rest received chemotherapy and/or granulocyte colony-stimulating factor mobilized peripheral blood stem cells. The harvest was unmanipulated in 28 patients, the rest receiving some form of lymphocyte depletion, mostly through CD34+ selection. Median followup was 16 months (range 3-55). Responses were measured using the American College of Rheumatology (ACR) criteria. Forty-nine patients (67%) achieved at least ACR 50% response at some point following transplant. There was a significant reduction in the level of disability measured by the HAQ (p < 0.005). Most patients restarted DMARD within 6 months for persistent or recurrent disease activity, which provided disease control in about half the cases. Response was significantly related to seronegative RA (p = 0.02) but not to duration of disease, number of previous DMARD, presence of HLA-DR4, or removal of lymphocytes from the graft. There was no direct transplant related mortality, although one patient, treated with the BuCYC regimen, died 5 months post-transplant from infection and incidental non-small cell lung cancer. CONCLUSION: Autologous HSCT is a relatively safe form of salvage treatment in severe, resistant RA. In these open label studies significant responses were achieved in most patients, with over 50% achieving an ACR 50 or more response at 12 months. Although the procedure is not curative, recurrent or persistent disease activity may be subsequently controlled in some patients with DMARD. Clinical trials are necessary to develop this approach in patients with aggressive disease who have failed conventional treatment including anti-tumor necrosis factor agents.