Three years ago we described a novel autocrine IL-3/G-CSF mechanism active in the leukemic CD34(+) cells from chronic myeloid leukemia (CML) patients in chronic phase (PNAS 96: 12804-12809, [1999]). We also showed that exposure of the most primitive CD34(+) cells from normal human bone marrow to excess IL-3 stimulates not only the division of these cells but also their differentiation. In contrast, both IL-3 and G-CSF cause an expansion of the more mature types of normal CD34(+) progenitors. These findings suggested that the autocrine IL-3/G-CSF mechanism active in CML stem cells can compromise their self-renewal in spite of increasing their proliferative activity, which, in turn, might explain the paradoxically slow rate of expansion of this compartment over time in patients with latent disease. To investigate this hypothesis, we have begun to characterize the numbers and types of cells generated from chronic phase CML patients' cells transplanted into adult immunodeficient mice or fetal sheep, and also from transplants of primitive murine and human hematopoietic cells transduced with a retroviral BCR-ABL vector. Our findings to date using these models reinforce the importance of the autocrine IL-3/G-CSF mechanism in the development of CML. BCR-ABL appears to directly activate IL-3 and G-CSF production in primitive hematopoietic cells and this is important to their transplantable leukemogenic activity. However, the development in vivo of an overt leukemia from primitive BCR-ABL(+) hematopoietic cells can be very delayed. These models thus offer new opportunities for analyzing the molecular events that underlie the pathogenesis of human CML and the future testing of new therapeutic approaches.