Currently available chemotherapy has probably reached the limits of its potential in treating acute myeloid leukemia (AML). In considering the next steps it is appropriate to exploit on the one hand knowledge of the molecular, immunophenotypic and biological characteristics of the disease and on the other the biology of the patient. The aim is to move towards a more targeted approach. Immunophenotyping has defined an adequate target (CD33) for antibody-directed treatment, although this is not leukemia specific. Monotherapy has produced important response rates in relapsed disease but it is unlikely to displace conventional chemotherapy. Several randomized trials of antibody directed chemotherapy in combination with chemotherapy nearing completion will establish the usefulness of this approach. In most patients a leukemia-specific immunophenotype can be characterized that can be used to monitor treatment. Minimal residual disease (MRD) detection in morphological remission can detect patients at high risk of relapse, as can a limited number of molecular markers. The clinical value of intervening at the time of MRD detection is not clear. Among the increasing molecular abnormalities described in AML, FLT-3 mutations appear the most attractive for therapeutic intervention. Several phase 2 studies have shown limited efficacy, and randomized trials in combination are underway. Other mechanisms that can be specifically targeted include farnesylation, methylation status, and histone deacelylation. Newer knowledge about the immunophenotypic and biological characteristics of the leukemic stem cell population has opened opportunities to develop treatments that exploit characteristics of the leukemic stem cells that differ from the normal stem cell. Some of these initiatives are now discussed.