Between 1978 and 1983, 1127 patients with de-novo acute myeloid leukaemia (AML) were entered into the Medical Research Council (MRC)'s 8th AML trial. All received the same induction therapy consisting of daunorubicin, cytarabine, and 6-thioguanine--DAT (1 + 5). The 67% who entered complete remission were randomised to consolidation with two or six further courses of DAT. Adults under the age of 55 were randomised for central nervous system (CNS) prophylaxis with intrathecal cytarabine and methotrexate. Finally, those still in remission after 1 year of cytarabine and 6-thioguanine (AT) maintenance were randomised to receive either late intensification with cyclophosphamide, vincristine, cytarabine, and prednisolone (COAP) or continued AT. The median survival for the whole group was 12 months; the median duration of first remission was 15 months, with relapse-free survival at 5 years estimated at 18%. The factors most strongly associated with poor survival were performance status and age at presentation, but even among those over 60 years of age, half went into remission. Six courses of DAT consolidation gave a small advantage over two courses in reducing the number of late relapses but no significant survival advantage. Late intensification showed a marginally significant advantage over continued AT maintenance. The incidence of CNS relapse was low and unaffected by prophylaxis. The second remission rate varied from 10% when the first remission was shorter than 6 months to 61% when it had continued for more than 2 years. 40 patients received histocompatible allogeneic bone-marrow transplants in first remission. There was a high procedure-related death rate, particularly among patients over 30 years of age. Thus, initially at least, the transplanted group had shorter survival than a comparable group of chemotherapy-treated patients. Treatment specifications remained unchanged throughout the trial but those enrolled in the later half of the trial had a better (p = 0.003) survival.