Over the last three decades there have been dramatic advances in deciphering the cytogenetic and molecular lesions underlying the pathogenesis of acute myeloid leukemia (AML). These have not only afforded greater insights into disease biology, but also provided useful information predicting the likelihood of any given patient achieving and maintaining remission following conventional chemotherapy, leading to the development of risk-stratified treatment approaches. However, it is becoming increasingly apparent that AML is highly heterogeneous at the molecular level. Defining the individual genetic abnormalities or combinations of markers that provide significant independent prognostic information and establishing their respective relationships to other pre-treatment characteristics that impact on outcome, such as age and presenting white blood cell count, presents a major ongoing challenge. Moreover, there is increasing evidence that risk of relapse and overall survival can be predicted by assessment of kinetics and depth of response following front-line therapy and monitoring of the leukemic burden using molecular or immunological approaches to minimal residual disease (MRD) detection. These advances present the exciting prospect that panels of pre-treatment parameters affording independent prognostic information can be integrated with precise measurement of treatment response using MRD technologies to provide greater refinement in risk-adapted management of AML. This could lead to further improvements in outcome and serve to identify in a more reliable fashion those patients most likely to benefit from allogeneic transplant in first remission.