Therapy-related myelodysplasia and acute myeloid leukaemia (tMDS/tAML) are unfortunate consequences of chemotherapy. Risk factors for developing tMDS/tAML include increasing age and exposure to alkylating agents and fludarabine (Pedersen-Bjergaard & Andersen, 2001). The crude incidence of tMDS/tAML following fludarabine containing regimens has been estimated at 3·5–8% (Morrison et al, 2002; McLaughlin et al, 2005; Milligan et al, 2006). As the median age of patients in these trials was only about 58 years, we analysed the incidence of tMDS/tAML following fludarabine, in an older population attending a general clinic. Records of all patients with indolent lymphoproliferative disorders treated with fludarabine at our hospital between 1993 and 2004 were analysed to include fludarabine dose, all chemotherapy regimens and the presence of progressive severe cytopenias. Severe cytopenias were defined as Hb < 8 g/dl, neutrophil count < 1·0 × 109/l, plateletcount < 50 × 109/l that persisted for more than 10 weeks from chemotherapy. The bone marrow spreads and trephine biopsies of patients with cytopenias were examined. If the cytopenias were not because of marrow infiltration from the lymphoproliferative disease, the marrows were examined independently by three haematologists. Differences between patient groups were evaluated using the chi-squared test for categorical data, and the Mann–Whitney test for continuous data. Fifty-seven patients were identified (follicular lymphoma, 24 patients; chronic lymphocytic leukaemia, 24 patients and lymphoplasmacytoid lymphoma, nine patients). Four patients were excluded from further analysis (immediate bone marrow transplant, two patients; lost to follow up, one patient and non-compliance, one patient). We excluded the 12 patients who died within 1 year of the first dose of fludarabine (resistant disease, eight patients; cerebrovascular accident, two patients; carcinoma lung, one patient and pancreas, one patient). We therefore analysed 41 patient records (median age 72 years) for the long-term effects of fludarabine. Initially, patients received fludarabine alone, but subsequently patients received fludarabine with cyclophosphamide (cyclophosphamide 10 mg/1 mg of fludarabine). Therapy-related myelodysplasia and acute myeloid leukaemia developed in eight of the 41 patients; crude incidence 20%, actuarial incidence 33% (95% CI: 22–42%) at a median of 22 months (range 12–43 months) from the start of fludarabine treatment. All eight have died, six from tMDS/tAML (at a median of 2·5 months from diagnosis), one following stem cell transplant and one from pancreatic carcinoma. Four of these eight patients had tMDS/tAML diagnosed before death, but four cases were only diagnosed retrospectively on bone marrows performed for progressive cytopenias. These four died with profound neutropenia (range 0·03–0·3×109/l) and three of four were platelet transfusion dependent. These four marrows had originally been interpreted as showing; immune neutropenia with positive red cell direct antiglobulin tests (DAGT), two patients, progressive disease, one patient and therapy-induced hypocellular marrow with positive DAGT, one patient. The median dose of fludarabine was significantly higher in the tMDS/tAML group than in the non-tMDS/tAML group (P = 0·01, see Table I) All patients developing tMDS/tAML had received fludarabine and cyclophosphamide as had 25 of 33 of the non-tMDS/tAML patients (see Table I). None of the seven patients who received fludarabine alone developed severe cytopenias and tMDS/tAML (their marrow samples were re-reviewed). All tMDS/tAML patients received prior alkylating agents as did 29 of 33 of the non-tMDS/tAML patients. This study found a higher incidence of tMDS/tAML following fludarabine than previously reported. This may be due to the older age of our patients, as the risk of developing tMDS/tAML may increase proportionally to the square of patient age (Pedersen-Bjergaard et al, 1987). All patients who developed tMDS/tAML received fludarabine with cyclophosphamide. Cytogenetic abnormalities in tMDS/tAML following fludarabine are similar to those following alkylating agents (Morrison et al, 2002; McLaughlin et al, 2005). Therefore, the combination of fludarabine with an alkylating agent, such as cyclophosphamide, might potentiate the damage. It is known that patients given two alkylating agents together have a very high risk of developing tMDS/tAML (Pedersen-Bjergaard et al, 1987). Fludarabine is also immunosuppressive, thus possibly reducing the immune surveillance necessary to prevent leukaemias. We initially aimed to identify the denominator for the four cases of tMDS/tAML who were already known. During analysis we noted four patients with progressive cytopenias and, after re-examining the marrows, found unrecognised tMDS/tAML. Similarly, Bennett et al (2005) looked for tMDS/tAML following iodine I131 tositumomab and found that 40% of their cases had had unrecognised tMDS prior to I131 tositumomab. Although the tMDS/tAML patients may have been a poor prognosis group, those who died of tMDS/tAML had controlled underlying lymphoproliferative disease. Our high incidence of tMDS/tAML may be due to the older age of our patients. Our data suggest that the risk of tMDS/tAML may be related to fludarabine dose, and that adding cyclophosphamide may possibly enhance this risk. tMDS/tAML may be under-recognised, especially in the elderly.