Rituximab (anti-CD20 monoclonal antibody) represents a new interesting tool for treating B-cell malignancies, especially low-grade lymphomas (McLaughlin et al, 1998). To our knowledge, only one paper has reported on the efficacy of this agent for treating cold agglutinins and refractory autoimmune haemolytic anaemia (AIHA) complicating B non-Hodgkin's lymphoma (Lee & Kueck, 1998). We would add a similar clinical case in which this monoclonal antibody has played a significant role. A 73-year-old Caucasian man was admitted to the Haematology Department to December 1997 because of fatigue and anaemia. His past medical history was unremarkable. Physical examination found the patient to be icteric and he did not have hepatosplenomegaly or lymphadenopathy. Haemoglobin (Hb) was 8·5 g/dl, reticulocytes were 190 × 109/l, haematocrit was 25%, mean corpuscular volume (MCV) was 104 fl, platelets were 383 × 109/l, white blood cells were 6·3 × 109/l including 58% neutrophils, 1% eosinophils, 1% basophils, 33% lymphocytes and 7% monocytes. Biochemical evidence of haemolysis was supported by the following results: increased bilirubin at 51 μmol/l, lactate dehydrogenase (LDH) at 495 IU/l (normal < 450 IU/l) and haptoglobin below the detection limit. Serum protein electrophoresis and immunofixation were normal. Immunoglobulin (Ig) levels were: IgG 11·7 g/l, IgA 4·86 g/l and IgM 1·14 g/l. The direct anti-globulin test was positive for polyvalent serum (++++), complement (++++) and IgG (weak). Cold agglutinins of the IgM type, showing anti-I specificity, were strongly positive with a titre of 1/4096. Serologies for human immunodeficiency virus (HIV), hepatitis B and C, syphilis and mycoplasma were negative. A nodular infiltration by CD20-positive lymphoplasmacytic cells was evidenced at the trephine biopsy. This cell population was not detected in peripheral blood on morphological or immunophenotypical investigations. Beta-2 microglobulin was 3 mg/l (normal < 2 mg/l). No lymphadenopathy was seen on thoracic and abdominal computerized tomography (CT) scans. This picture was suggestive of bone marrow-localized low-grade B non-Hodgkin's lymphoma. The patient (body weight: 60 kg) was initially treated with oral cyclophosphamide (100 mg/d) and prednisolone (1 mg/kg/d) plus folic acid. Four weeks after introduction of therapy, his Hb was about 10 g/dl. As soon as steroids were decreased, haemolytic signs reappeared. By May 1998, Hb was 12·7 g/dl with 60 mg/d prednisolone alone. Again, steroids were tapered until 25 mg/d in January 1999. During this period, Hb continued to fall below 10 g/dl. In spite of increased dosage of steroids up to 70 mg/d, Hb levels did not improve. Then, in May 1999, the patient was placed under cyclosporin up to 400 mg/d with reduced doses of prednisolone. Because of lack of efficacy and significant renal toxicity, this treatment was stopped in July 1999. Then, five monthly intravenous boluses of cyclophosphamide were given without any change of transfusional needs. The patient refused to receive polychemotherapy and was under high-dose oral prednisolone (100 mg/d). Three therapeutic plasma exchanges were performed in November 1999. All these modalities did not induce a significant effect on Hb levels and transfusion requirements. This man demonstrated a high degree of steroid-related toxicities including cushingoid facies, myopathy and osteoporosis. Furthermore, by December 1999, 33 units of red cells had been transfused since diagnosis, inducing a marked haemochromatosis picture (serum ferritin, 5093 μg/l; normal, 19–495 μg/l). Considering the experience of Lee & Kueck (1998), our patient was treated from mid-December 1999 with four weekly injections of rituximab at 375 mg/m2 (Mabthera®, Laboratoires Roche, Neuilly, France). No side-effects were noted. Red cell transfusions were stopped after the second rituximab infusion and the Hb level raised above 10 g/dl in January 2000. Under this therapy, at the last evaluation performed 9 months later, this man remained transfusion free and demonstrated a Hb level of 12·5 g/dl with normal bilirubin, LDH and haptoglobin levels. Nevertheless, the direct anti-globulin test remained positive. No haemolysis relapse has been observed to date in spite of the tapering and arrest of corticosteroids. As in the case report of Lee & Kueck (1998), our patient had a marrow-localized CD20-positive chronic lymphoproliferative disease, cold agglutinins and a severe AIHA. This latter manifestation was unresponsive to steroids (albeit an initial response was obtained), cyclophosphamide, cyclosporine or plasma exchange. It is clear that, in this case, rituximab was active against the haemolytic phenomenon. The tolerance was excellent. Interestingly, the response was relatively rapid and no evidence of relapse has been noted 9 months after therapy initiation. Clinicians must be aware of this therapy tool in such severe situations. Reporting of further cases may provide important information regarding the management of refractory AIHA in this setting.