359 eligible children with acute myeloid leukaemia (AML) entered the MRC AML 10 trial between May 1988 and March 1995. Patients received four courses of intensive induction and consolidation chemotherapy, with or without subsequent autologous (A-BMT) or allogeneic (allo-BMT) bone marrow transplant. There were randomized comparisons of thioguanine versus etoposide in induction and of A-BMT versus not. Allo-BMT was recommended for patients with a HLA-matched sibling and was evaluated by donor versus no donor comparison. The complete remission rate was 92%. In first remission there were 20 deaths during consolidation chemotherapy and 11 after BMT (8/61 allo-BMTs. 1/60 A-BMTs and 2/4 matched unrelated donor transplants). The relapse rate was low, decreasing from 26% in the first year to 2% in the fourth. Long-term outcome was excellent with survival at 7 years from entry of 56% and event-free survival of 48%. There were no significant differences between thioguanine and etoposide, whereas both A-BMT and allo-BMT reduced relapse risk but did not produce a significant survival benefit. It appears that over half the children entered into AML 10 are cured, a result which compares favourably with other reported series. We conclude that four courses of intensive chemotherapy are an effective approach to the treatment of paediatric AML, which avoids the acute toxicity and long-term side-effects of BMT and also avoids the need for prolonged maintenance therapy or cranial irradiation.