The lympho-hematopoietic colony-forming assay has been redesigned into a rapid, nonsubjective and standardized proliferation assay that can measure the effects of compounds on multiple stem and progenitor cell populations from different species simultaneously using a sensitive, high-throughput bioluminescence readout. Eleven reference compounds from the Registry of Cytotoxicity (RC) and eight other compounds, including anticancer drugs, were studied over an 8- to 9-log dose range for their effects on seven cell populations from both human and mouse bone marrow simultaneously. The cell populations studied included a primitive (HPP-SP) and mature (CFC-GEMM) stem cell, three hematopoietic (BFU-E, GM-CFC, Mk-CFC) and two lymphopoietic (T-CFC, B-CFC) populations. The results reveal a five-point prediction paradigm for lympho-hematotoxicity. Depending on how and which populations are affected, the resulting effects in the periphery can be predicted. Validation against the RC Prediction Model produces a high degree of correlation between the in vitro IC(50) values and known in vivo LD(50) values, thereby allowing preclinical dosing to be predicted. If primary human hematopoietic target tissue is used, inhibitory concentration (IC(50)/IC(75)/IC(90)) values of anticancer and other drugs can be converted into predicted clinical doses which, when compared to published chemotherapeutic dosing regimen, are very similar. When performed during early drug screening, the prediction value of the assay should help reduce time and cost, but above all, provide increase efficacy and safety for the patient.