Until recently, lineage fidelity was thought to be preserved in leukaemic cells, which by available tests showed surface markers and enzymatic patterns characteristic of an appropriate normal cell lineage and stage of differentiation. Our data indicate that this theory is too restrictive. If leukaemogenesis occurs in pluripotent progenitors in a relatively high percentage of cases, we would propose a model in which lymphoid and myeloid differentiation antigens are expressed simultaneously until the progenitor cell commits to a single lineage. Lineage commitment could involve external factors, e.g. growth factors (Sherr et al, 1985), that cause genes specific for the opposite lineage to be 'switched off'. The control of gene expression in mammalian cells and the specific chromosomal sites of genes coding for the various lineage-associated markers remain uncertain. However, recent studies indicate that most, if not all, leukaemic cells contain chromosomal abnormalities, many involving rearrangements of DNA (Williams et al, 1986). Since the control of eukaryotic gene expression is known to involve numerous sequence elements, some acting at a distance from the site of transcription (Dynan and Tjian, 1985), genetic perturbations within the cell (e.g. a reciprocal translocation) could be expected to deregulate certain genes, leading to their under- or overexpression analogous to activation of the c-myc oncogene by the 8;14 translocation in Burkitt's lymphoma. Thus, an almost infinite variety of cell lineage-related phenotypes could be expected from this mechanism alone, even if the transforming event did not involve a pluripotent stem cell. Also, we have hypothesized that enzymes such as TdT, a DNA polymerase that catalyses polymerization of deoxyribonucleotides without a DNA template, could serve as a modifier of DNA sequences, permitting otherwise inactive genes to be expressed (Stass and Mirro, 1985). It is interesting that most cases of childhood acute mixed-lineage leukaemia are TdT positive, even though this is not true for the chronic leukaemias of adults. It is now clear that unusual combinations of myeloid and lymphoid cell lineages are much more common in acute leukaemia than have been generally recognized or suspected. The traditional division of the acute leukaemias into ALL and AML may not be the most accurate way to represent this class of haematological malignancies. That mixed-lineage leukaemia may require alternative therapy is a clinically important observation and underscores the need for comprehensive testing of blast cells at diagnosis.(ABSTRACT TRUNCATED AT 400 WORDS)