The c-fms proto-oncogene encodes an integral transmembrane glycoprotein with tyrosine specific protein kinase activity whose properties resemble those of receptors for polypeptide growth factors. The relatively restricted expression of the feline c-fms gene in mononuclear phagocytes (peripheral blood monocytes and tissue macrophages) and their committed bone marrow progenitors suggested that c-fms encoded a receptor for a macrophage specific growth factor. We found that the receptor for the mononuclear phagocyte colony stimulating factor, CSF-1 (M-CSF), is biochemically and immunologically related to the c-fms gene product, consistent with the hypothesis that c-fms is identical to the CSF-1 receptor gene (Sherr et al, 1985). Although the activity of CSF-1 has been defined through its action on haemopoietic cells, the c-fms gene is expressed in human placenta (Müller et al, 1983a) and in human choriocarcinoma cell lines derived from placental trophoblasts (Müller et al, 1983b). The latter cell lines exhibit binding sites for CSF-1, suggesting that this colony stimulating factor might also have an embryological role in placental development. The retroviral oncogene v-fms is a potent fibroblast transforming gene, whereas c-fms can be expressed at relatively high levels in certain normal tissues without causing neoplastic transformation. The glycoprotein encoded by v-fms is closely related to the c-fms gene product but differs at its extreme carboxy terminal end. Because v-fms, like c-fms, encodes a competent ligand binding domain, cells producing the v-fms gene product acquire the ability to bind CSF-1. Moreover, many fibroblast cell lines susceptible to transformation by v-fms produce the growth factor. Although this raises the possibility that v-fms transforms cells by an autocrine mechanism, antibodies to epitopes in the v-fms-coded ligand binding domain that interfere with CSF-1 binding, or antibodies to CSF-1 itself, do not affect the transformed phenotype. In membrane preparations, tyrosine-specific phosphorylation of the v-fms product appears to be constitutive, whereas in vitro phosphorylation of the c-fms-coded glycoprotein on tyrosine is enhanced in the presence of CSF-1. These results are most compatible with the possibility that critical alterations in the 3' coding region of the c-fms gene activate its kinase activity and unmask its latent transforming potential. An implication of these findings is that chromosomal rearrangements affecting c-fms could contribute to myeloid leukaemogenesis.