To understand the origins, and disease progression, of leukaemia we first need a clear idea of how the progeny of haematopoietic stem/precursor cells normally choose their fates. For about 30 years, 'classical' models of blood cell development have envisaged a branching tree with two trunks representing the two major families of cells: myeloid/erythroid and lymphoid. Recent debate about this apparent dichotomy has given rise to new models of haematopoiesis and new ways of viewing stem-cell behaviour. These suggest that stem and progenitor cells are more versatile than was first appreciated, so there can be multiple routes to one type of end cell. An important aspect of this versatility during haematopoiesis is that progenitor cells retain an unexpected portfolio of clandestine lineage potentials even when they seem to have progressed quite far along a particular developmental pathway. Here we examine this decision-making process and ask whether, developmentally, leukaemia stem cells are equally or less versatile than their normal counterparts.