The aim of the present study was to analyze whether or not leukemic clonogenic cells are restricted to the CD34+ cell fraction and to investigate the effect of IL-3 and G-CSF on blast cell populations dissected according to their CD34 reactivity. For this purpose 34 patients were studied. Patients were classified into three groups according to CD34 antigen expression: (1) cases in which all blast cells (100%) were positive for the CD34 Ag (n = 9); (2) cases in which all blast cells lacked the expression of this antigen (n = 10); and (3) patients in whom both, CD34 positive and negative blast cell subsets coexisted (n = 15). In 15 cases immunomagnetic cell selection was performed and two subpopulations were separated: one, phenotypically more immature (CD34+), and another, theoretically more differentiated (CD34-/33+). In addition, in three cases both CD34+ and CD34- blast cell subpopulations were sorted using a FACStar flow cytometer. Blast colony assays were performed using 0.9% methylcellulose and two different recombinant human hematopoietic growth factors (HGFs), IL-3 and G-CSF, were used as growth stimulants. Either, a single or a combination of the growth factors was added to cultures. Colony formation was observed in both 100% positive or 100% negative cases for the CD34 antigen as well as in the CD34+ and CD34- cell fractions separated by immunomagnetic selection or flow cytometry. The effect of G-CSF and IL-3 on both cell fractions was as follows: cases with a uniform population according to CD34 expression (100% positive or negative) showed a better growth response with IL-3 especially for the CD34+ cases (87% vs 40% of CD34+ and CD34- cases, respectively). Within the CD34-/33+ selected fractions, IL-3 tended to induce a higher proliferative response than G-CSF while the opposite was found within the CD34+ cell selected fractions. In contrast it was observed that both IL-3 and G-CSF induced a higher PE on the CD34- blast cells (both selected and 100% negative), although the difference was not statistically significant. The existence of a possible synergistic effect (SE) between HGFs was also explored. Overall, a synergistic growth was observed in nine out of the 13 selected cases studied and this effect could be seen in both CD34- or CD34+ blast cell fractions. The analysis of the complete phenotypic characteristics of these cells revealed that cell fractions showing SE were more immature according to the expression of CD15 and HLA-DR antigens. We can conclude that in leukemic hematopoiesis, CD34 antigen expression does not have the same significance as it does in normal hematopoiesis since clonogenic cells are not restricted to the CD34+ acute myeloid leukemia (AML) blast cell fraction. Moreover, our study shows that the heterogeneous response to HGFs observed in AML patients may be associated with the existence of immunophenotypically different blast cell subsets.