The pretreatment karyotype of leukemic blasts is currently the key determinant in therapy decision making in acute myeloid leukemia (AML). The World Health Organization (WHO) has recognized this important information by including, besides clinical, cytological, cytochemical, and immunophenotypical features, recurrent cytogenetic abnormalities in its classification (Table 1). However, although the WHO defines important biologically and clinically relevant entities, the prognostic value of some of the well-defined cytogenetic subgroups is partially masked in the WHO classification. Moreover, in the recent past a number of novel molecular aberrations with marked prognostic value, which are not yet incorporated in the WHO classifications have been identified. These molecular abnormalities include mutations (e.g., in FLT3, c-KIT, and NPM1), partial duplications (e.g., of MLL and FLT3), and abnormal expression of pathogenetic genes (e.g., EVI1, WT1, BCL2, MDR1, BAALC, and ERG). In addition, novel molecular approaches in genomics, like monitoring the expression levels of thousands of genes in parallel using DNA microarray technology, open possibilities for further refinement of prognostication of AML. Gene expression profiling in AML is already well established and has proven to be valuable to recognize various cytogenetic subtypes, discover novel AML subclasses, and predict clinical outcome. The current advances made in molecular understanding of AML will ultimately lead to a further refinement of prognostics of AML.