CD7 has been used as a valuable marker for normal and malignant T cells and also for a proportion of acute nonlymphocytic leukemia (ANLL) cells. Difference in reactivity was noticed among CD7 antibodies, however, when tested against ANLL cells and myeloid/monocytoid cell lines; Tp40 antibody produced in our laboratories was not reactive with the HL-60 promyelocytic line, whereas 4A antibody was reactive, even though both detected a quite similar or an identical epitope on CD7 molecule. Preincubation of HL-60 cells with human immunoglobulin preparation clearly negated the reactivity by 4A, suggesting that 4A antibody is not reactive to CD7 itself, but it probably binds with immunoglobulin G Fc receptors expressed on HL-60 cells. Five cases of ANLL which were positive with 4A antibody were selected and tested with Tp40 antibody, and only two were found to be positive. Expression of CD7 mRNA in these two cases (but not in other cases) was also demonstrated by Northern blotting with a cDNA probe for CD7 recently cloned in our laboratories, indicating that CD7 is expressed on a certain fraction of ANLL, although the positive cases may be smaller than the reports so far appeared. A Northern blot study was also conducted with two acute lymphocytic leukemia cases and one lymphoblastic lymphoma case with CD7+, CD2-/+/-, CD5-/+/- phenotype and germline T cell receptor beta genes. CD7 mRNA is expressed in all three cases and CD3 mRNA is also observed in two cases, suggesting that these tumor cells are of T precursor origin.