We herein describe an unusual case of acute myeloid leukaemia (AML) showing strong cytochemical reactivity for myeloperoxidase (MPO) but surprisingly no reactivity using flow cytometry for any of the lineage-specific cell surface markers, i.e. myelomonocytic antigens CD13, CD14 and CD33; or B-lymphoid antigens CD19, CD20 and immunoglobulins; or T-lymphoid antigens CD2, CD3 and CD5. The strong reactivity for MPO and the complete absence of reactivity for CD13 and CD14 was verified by an independent assay involving alkaline phosphatase-anti-alkaline phosphatase (APAAP). Our case is of interest for at least two reasons: First, a poorly differentiated variant of AML (negative for MPO but positive for one or more of the myeloid-lineage CD antigens) has been designated FAB M0. In terms of the expression of phenotypic markers, our case may be considered as an 'MPO (+), CD antigen (-) AML'. The CD antigens are known to be expressed very early during myeloid differentiation whereas MPO (in its functional form) is viewed as being expressed relatively late in the process. It is therefore intriguing from a biological standpoint why the supposedly early antigens (CD33 and CD13) remain unexpressed; this may represent an example of 'asynchronous differentiation' in leukaemia. Second, from a practical standpoint, the use of immunophenotyping as a first-line diagnosis would fail to detect such cases. This case strengthens the notion that immunophenotyping by flow cytometry does not eliminate the necessity of performing peroxidase cytochemical staining.