We read with interest the case report of a patient with Jo-1-positive polymyositis treated effectively with autologous stem cell transplantation (ASCT) (Baron et al, 2000) and would like to report our experience of ASCT in polymyositis. After an initial period of inflammatory polyarthritis, our patient (a 38-year-old woman) developed Jo-1-positive polymyositis [creatine kinase (CK) level 5754 IU/l (0–140)] with restrictive lung disease. Her disease progressed rapidly and was resistant to therapy with prednisolone (60 mg), azathioprine (150 mg), intravenous cyclophosphamide pulses (total 15 g), intravenous immunoglobulin, cyclosporine (150 mg) and plasmapheresis. In July 1997, stem cells were mobilized with cyclophosphamide (Cy) 2 g/m2 and granulocyte colony-stimulating factor (GCSF) 5 μg, and the graft underwent CD34-positive selection. After 6 g/m2 of Cy, 3·7 × 106 of CD34+ cells/kg and 4·04 × 104 of CD3+ cells/kg were reinfused. The transplant proceeded without any serious complications. During the first 3 weeks the patient felt much better with increased muscle strength and stamina; however, relapse occurred on d 21 (CK 241 IU/l). Despite recommencing methotrexate and cyclosporine and increasing prednisolone to 35 mg, her muscle strength deteriorated (CK 110–780 IU/l) and she required plasmapheresis in January 1998. In early 1999 she developed two subcutaneous abscesses and aspiration revealed Mycobacterium avium complex. Peripheral blood FACS analysis revealed that CD3, CD4 and CD8 T-cell levels, and NK and B-cell levels were all low (0·06 × 109/l, 0·03 × 109/l, 0·05 × 109/l, 0·10 × 109/l and 0·01 × 109/l respectively). Her disease stabilized for the next few months; methotrexate and cyclosporine were discontinued and prednisolone reduced to 12·5 mg/d. Lung function tests showed no progression over the preceding 12 months. For the last year, her CK has remained stable (250 IU/l) but there has been a mild clinical deterioration. This patient underwent ASCT for aggressive polymyositis with rapidly progressive lung and muscle disease and, although she felt clinically better for only 3 weeks, the rate of progression of her disease has clearly slowed. She has subsequently responded to previously ineffective medication, suggesting an immunomodulatory effect. However, our patient did not respond as dramatically as the patient reported by Baron et al, 2000 and unfortunately suffered an opportunistic infection 2 years post transplant. Some variation in outcome may be caused by differences in the method of ASCT used in these two patients. Our patient received a less intensive conditioning regime, which may account for her reduced response. Both patients received manipulated grafts and our patient received a smaller number of CD3+ cells back. The case reported by Baron et al (2000) was shown to have peripheral lymphocyte depletion for the first 6 months post ASCT, but our patient continued to have impaired reconstitution at 20 months. This probably contributed to opportunistic infection. Our patient had also received a large amount of immunosuppression before the transplant, which may have impaired her ability to reconstitute. The use of ASCT in autoimmune disease is still in its exploratory phase (Bingham et al, 2000). A balance must be struck between the wish to give patients a large amount of conditioning in order to achieve complete myeloablation and the need to minimize transplant-related toxicity. This point is borne out by the lack of serious toxicity seen in six patients with rheumatoid arthritis (RA) transplanted in our unit (conditioned with Cy 200 mg/kg) (Bingham et al, 2001) and the fact that the only patient with RA who died after ASCT had received BuCy (Durez et al, 1998, 1999; Tyndall et al, 1999).