An increasing number of reports document instances in which individual leukemic cells coexpress markers normally believed to be restricted to a single lineage. This has been interpreted by McCulloch and colleagues as aberrant programming or lineage infidelity and contrasts with earlier suggestions that lineage fidelity of gene expression was usually maintained in leukemia. We argue that several examples of infidelity are suspect on technical grounds, whereas others are bona fide and require explanation, eg, partial rearrangements and expression of Ig heavy-chain and/or T cell receptor genes in inappropriate cells and terminal deoxynucleotidyl transferase in leukemic myeloblasts. Individual examples of truly aberrant gene expression may well occur in leukemia but with insufficient regularity to be of general significance. We suggest that verifiable and consistent examples of apparent lineage infidelity do not reflect genetic misprogramming but rather the existence of a transient phase of limited promiscuity of gene expression occurring in normal biopotential or multipotential progenitors and able to be preserved as a relic in leukemic blast cell populations that are in maturation arrest. This alternative explanation has interesting implications for mechanisms of hematopoietic differentiation and leads to some testable predictions.