One hundred and fifty acute myeloid leukaemia (AML) patients in first remission received an allogeneic bone marrow transplant (BMT), after conditioning with cyclophosphamide 120 mg/kg and total body irradiation (TBI) 3.3 Gy x 3 (total nominal dose 9.9). The received dose, as recorded by thermoluminescent dosimeters, ranged between 7. 83 and 12.25 Gy. Patients who received TBI < 9.9 Gy (n = 34) had a significantly higher relapse rate when compared with patients receiving >/= 9.9 Gy (n = 116) (43% vs. 19%; P = 0.002). Graft versus host disease (GvHD) prophylaxis consisted of cyclosporin A (CyA) with or without methotrexate (MTX). The dose of CyA was either 1 or 5 mg/kg/day i.v. from day -1 to + 20, then 10 mg/kg/day orally until day + 365. Patients receiving 5 mg/kg CyA (n = 40) had a higher risk of relapse (49% vs. 15%; P = 0.0001). Thus, low-dose TBI (< 9.9 Gy) and high-dose CyA (5 mg/kg) were significant predictors of leukaemia relapse. Patients were then divided into three groups: those who had both negative predictors (< 9.9 Gy TBI and 5 mg/kg CyA; n = 26); those who had only one (either < 9.9 Gy TBI or 5 mg/kg CyA; n = 22); and those who had neither (>/= 9.9 Gy TBI and 1 mg/kg CyA; n = 102). The three groups were comparable for FAB subtype, interval diagnosis transplant and age. The 5-year actuarial relapse rate for these three groups of patients was 49%, 41% and 15%, with no difference between the first two and a significant difference when compared with the latter (P < 0.01). These data indicate that acute myeloid leukaemia can be cured with allogeneic bone marrow transplantation given an intensive conditioning regimen and low-dose immunosuppression post-graft. Either alone is insufficient to produce long-term disease-free survival. These results may be relevant for programmes of reduced intensity conditioning designed for patients with acute leukaemia.