Terminal differentiation of human promyelocytic leukemia (HL-60) cells can be induced by a variety of chemical agents and this process can be monitored readily by the generation of morphologically, histochemically, and functionally mature granulocytes and monocytes/macrophages. The availability of this model has heightened interest in the possible therapeutic role of inducers of myeloid differentiation for the treatment of leukemia and other neoplasms. In addition, however, potent cancer chemopreventive agents induce HL-60 cell differentiation at very low dose levels. Thus, as part of our search for natural product chemopreventive agents, extracts derived from nearly 400 plants were tested for their potential to induce HL-60 cell differentiation. As a result, 17 plant extracts were judged to be active (ED50 values < or = 4 micrograms/ml). One of most potent leads was an extract derived from Dirca occidentalis Gray (Thymelaeaceae) (ED50, 0.14 micrograms/ml), and bioassay-guided fractionation led to the identification of genkwanin (I), (+/-)-lariciresinol (II) and sitoindoside II (IV) as active principles, with ED50 values of 18.3, 1.1 and 0.069 microM, respectively. Based on these data, we conclude that the HL-60 cell differentiation system is a valid and useful model for the discovery of natural product cancer chemopreventive or chemotherapeutic agents.