R etinoids were studied both to identify what skeletal components are important in the modulation of normal and leukemic human myeloid clonal proliferation and differentiation in vitro and to elucidate the mechanism by which retinoids modulate proliferation of hematopoietic cells. Retinoids with a derivatized terminal carboxyl group were significantly less active than all-trans-retinoic acid, and those with the addition of two methyl groups to the cyclohexenyl ring of retinoic acid or substitution of its beta-cyclogeranylidene group with a 1,1,3,3-5-indanyl ring system were markedly more active than all-trans-retinoic acid. Five of the retinoids strongly inhibited clonal growth of the HL-60 and KG-1 human leukemic cell lines (50% inhibition in the range of 3 X 10(-10)-1 X 10(-8) M) and markedly stimulated normal human myeloid colony formation (granulocyte-macrophage colony-forming cells [GM-CFC] 150% stimulation in the range of 3 X 10(-9)-3 X 10(-8) M). Further studies suggested that: Common structural requirements of the retinoids were important in the modulation of both normal and leukemic hematopoiesis. The retinoids were able to inhibit leukemic proliferation without induction of differentiation of the neoplastic cells. Studies on normal human GM-CFC suggested that the retinoids did not act by themselves as a colony-stimulating factor (CSF), or by stimulating accessory cells to produce CSF, but either required earlier progenitor cells to become GM-CFC or enhanced the sensitivity of GM-CFC to the action of CSF.