The availability of an in vitro assay able to detect hematopoietic progenitor cells closely related to those responsible for marrow engraftment following autologous bone marrow transplantation (ABMT) prompted us to establish a procedure aimed at maximally increasing the concentration of the cyclophosphamide derivative mafosfamide used for marrow purging. It, therefore, was the aim of the present study to investigate in a group of patients with acute nonlymphoblastic leukemia (ANLL; n = 19) and acute lymphoblastic leukemia (ALL; n = 19) in complete remission the effect of mafosfamide at the level of adherent blast colony-forming units (blast colony-forming units, CFU-Blast), as well as multipotential (granulocyte erythrocyte macrophage megakaryocyte colony-forming units, CFU-GEMM), erythroid (erythroid burst-forming units, BFU-E), and granulocyte-macrophage (granulocyte-macrophage colony-forming units, CFU-GM) progenitor cells. When nonadherent marrow mononuclear cells (MNCs) were incubated (30 min, 37 degrees C) with increasing doses of mafosfamide (30-120 micrograms/ml), a statistically significant (p less than or equal to 0.0005) dose-dependent suppression of CFU-Blast growth was observed. The mean (+/- 1 standard error of the mean [SEM]) values of 50% inhibition (ID50) of the CFU-Blast growth were not significantly different for ANLL (106 +/- 5) and ALL (107 +/- 5) patients. Analysis of CFU-Blast ID50 distribution demonstrated that ID50 ranged from 100 to 120 micrograms/ml in 17 cases (45%), whereas it ranged from 60 to 100 micrograms/ml in 12 cases and from 120 to 160 micrograms/ml in 9 cases. A statistically significant (p less than or equal to 0.05), dose-dependent suppression of colony growth from multi-potential and lineage-restricted progenitor cells was also observed. However, the value of CFU-Blast ID50 was significantly higher (p less than or equal to 0.05) than CFU-GEMM, BFU-E, and CFU-GM ID50 and ID95 values. In conclusion, our data demonstrate that: 1) the CFU-Blast assay allows to detect on an individual basis the doses of mafosfamide used for marrow purging, and 2) the concentrations of mafosfamide extrapolated by using the CFU-Blast assay are significantly higher than those obtained with the CFU-GM assay. The absence of any detrimental effect on marrow engraftment in vivo supports the safety of the CFU-Blast assay to evaluate the dose of mafosfamide used for marrow purging before ABMT.