Acute myeloid leukaemia (AML) is a highly heterogeneous disease, with biologically and prognostically different subtypes distinguished by cytogenetic and molecular genetic analysis, as recognised in the latest (2008) WHO classification system. However, since the publication of this schema, application of various high throughout technologies including whole genome and exome sequencing of AML cases has revealed a plethora of recurrent mutational targets, including a number of genes encoding transcriptional regulators, not previously implicated in leukaemogenesis. Deciphering the combinations of mutations that cooperate to induce AML and determining which particular alterations (or combinations) confer independent prognostic information represent major ongoing challenges, which will necessitate analysis of large cohort sizes involving international cooperation. However, the uncertainty concerning optimal risk-stratification of AML based on the rapidly evolving picture emerging from molecular genetic profiling provides a strong rationale for evaluation of minimal residual disease (MRD) detection as a tool to refine outcome prediction. This modality may not only capture differences in treatment response that reflect the underlying molecular heterogeneity, but also inter-patient variability in drug availability and metabolism, which may also significantly influence outcome. Developments in leukaemia diagnostics coupled with the possibility that MRD may be tracked using molecular and/or flow cytometry-based methods in the majority of patients hold considerable promise to inform more personalized approaches to therapy.