The patterns of genetic change, clonal evolution, natural history and latency are very different in the paediatric leukaemias compared with adult epithelial cancers but are similar to those in other childhood cancers of mesenchymal stem cell origin. This distinction has a biological logic in the context of the selective pressures for clonal emergence in different developmental and cellular contexts and has a major impact on curability. Most childhood leukaemias and some other mesenchymal stem cell tumours are of fetal origin and can metastasize without corruption of restraints on cell proliferation or bypassing apoptosis. In marked contrast to most invasive or metastatic epithelial carcinomas in adults, these former cancers then retain sensitivity to therapeutic apoptosis. Moreover, their abbreviated and less complex evolutionary status is associated with less genetic diversity and instability, minimising opportunity for clonal selection for resistance. A minority of leukaemias in children and a higher fraction in adults do, however, have genetic alterations that bypass cell cycle controls and apoptosis imposition. These are the 'bad news' genotypes. The cellular and molecular diversity of acute leukaemia impacts also on aetiology. Paediatric acute leukaemias can be initiated prenatally by illegitimate recombination and fusion gene formation in fetal haemopoiesis. For acute lymphoblastic leukaemia (ALL) in children, twin studies suggest that a secondary postnatal molecular event is also required. This may be promoted by an abnormal or delayed response to common infections. Even for a classic case of a cancer that is intrinsically curable by systematic chemotherapy i.e. childhood ALL, prevention may turn out to be the preferred option.