The National Cancer Institute (NCI) sponsored a workshop to develop a set of standardized diagnostic and response criteria for acute myeloid leukemia (AML) clinical trials. The French-American-British (FAB) classification was retained for diagnosing AML, with the addition of patients with bone marrow morphologic features of a myelodysplastic syndrome and less than 30% bone marrow blasts, but with greater than or equal to 30% blasts in the peripheral blood. In this report, there are four important subgroups of AML not defined in the FAB classification that are discussed: undifferentiated acute leukemia, MO (AML lacking definitive myeloid differentiation by morphology or conventional cytochemistry but with ultrastructural or immunophenotypic evidence for AML), mixed lineage leukemia, and hypocellular AML. Definitions of response for clinical trials are presented to facilitate comparisons among different studies. Complete remission is considered the only response worth reporting in phase III trials, since lesser responses do not improve survival. Partial remissions may be of interest to identify active new agents in phase I and II studies. Monoclonal antibodies and cytogenetic studies are not part of the routine assessment of remission or reassessment at relapse, and their role in the evaluation of patients with AML is currently being evaluated in clinical trials. Although we recognize that some of the definitions in this report are arbitrary, generalized use of these guidelines will make results of clinical trials more comparable and interpretable.