Many questions about the biology of AML remain to be answered. The initial genetic lesions that inhibit differentiation and increase the likelihood of self renewal have yet to be identified. Given the heterogeneity of this neoplasm, it is possible that many different mutational events may be capable of triggering leukemia. Alternatively, there may be only a small number of possible initial leukemic mutations, and the heterogeneous phenotype of the disease is determined by the evolution of different subclones that have acquired different secondary mutations. Studies with retroviral oncogenes have suggested that a common secondary event in an evolving myeloid tumor is the development of growth factor independence by either leukemic cell production of CSF or possibly constitutive activation of a CSF receptor. These mechanisms have not yet been established as important in human AML, although there is intriguing evidence to suggest that CSF genes are inappropriately activated in many cases.