A systematic review of chemotherapy trials in several tumour types was performed by The Swedish Council of Technology Assessment in Health Care (SBU). The procedures for the evaluation of the scientific literature are described separately (Acta Oncol 2001; 40: 155-65). This synthesis of the literature on chemotherapy in patients with acute myeloid leukaemia (AML) is based on 129 scientific articles: one meta-analysis, 51 randomised trials, 39 prospective and 18 retrospective studies, and 20 other articles. Altogether, 39,557 patients were included in these studies. The conclusions reached can be summarized into the following points: Standard induction therapy for patients with AML, consisting of daunorubicin and ara-C in conventional doses, results in a complete remission (CR) rate of 50-60% in an unselected population and a long-term survival of about 10-20%. The total doses of both ara-C and daunorubicin are of importance for remission duration and in some studies also for survival. High-dose ara-C in the induction therapy prolongs remission duration in randomised trials, but has not been proven to affect long-term survival. It also increases toxicity and is not generally recommended. Idarubicin, another anthracyclin, has been compared with daunorubicin in conjunction with ara-C, resulting in a higher CR rate, especially in younger patients. In a meta-analysis of the five-randomised trials performed, a slight survival advantage was also seen with idarubicin. Yet, there is inconclusive evidence to conclude that idarubicin is superior to daunorubicin, and further trials are needed. Mitoxantrone improves the outcome of induction therapy in comparison with daunorubicin in some randomised studies, but conclusive evidence is still lacking. The addition of etoposide to daunorubicin or mitoxantrone and ara-C has improved CR rates, but has not convincingly improved survival and secondary leukaemias may be induced. New induction treatment strategies are defined by identification of prognostic subgroups. A risk stratification of AML patients as to chromosomal aberrations might be of importance for the choice of therapy. Moreover, the speed and the morphological response to the first induction course are predictive for relapse. However, no prospective randomised studies are as yet published regarding risk-adapted induction therapy. Post-remission dose-intensive chemotherapy prolongs the duration of remission, seemingly most in patients < 60 years. However, the data in support of these conclusions are sparse. A convincing effect on survival has not been shown. Limited data indicate that post-remission maintenance therapy with long-term attenuated chemotherapy prolongs time to recurrence, without evidence for prolongation of survival. Allogeneic bone marrow transplantation is an established practice for consolidation in first remission for young patients with an HLA-matched sibling. It is however not known which patients will really benefit from transplantation as no truly randomised comparison of allogeneic vs autologous transplantation or conventionally-dosed chemotherapy has been performed. Patients with and without an HLA-identical sibling have been compared on the basis of intention-to-treat principles ('genetic randomisation'). The disease-free survival seems to be prolonged in the donor group, due to a lower relapse rate with allogeneic transplantation. A higher procedure-related mortality makes the effects on total survival uncertain. Randomised trials with autologous transplantation vs conventional consolidation show a lower relapse rate and a trend for an improved disease-free survival. In one study, in which an autograft was added to four courses of intensive therapy, there was also a late survival advantage. Thus, the role for intensified post-remission treatment in first complete remission with high-dose chemotherapy followed by allogeneic or autologous marrow or stem cell transplantation requires further studies. Moreover, studies with stratification of therapy according to predictors for prognosis in the individual patient are needed. Allogeneic stem cell transplantation after minimal or reduced myeloablative conditioning ('mini-transplantation' or non-myelo stem cell transplantation) induces a host-vs-graft tolerance and an immune graft-vs-leukaemia effect. This new concept of immunotherapy seems to have a low procedure-related mortality, but long-term effects are unknown and evaluation in controlled clinical studies is required. Patients with relapsed AML can only infrequently achieve long-term remissions with chemotherapy in conventional doses. trolled data indicate that allogeneic transplantation can be a curative treatment for these patients a