AIMS--To determine whether the use of different sources of anti-CD13 and anti-CD33 monoclonal antibodies leads to discrepant results in childhood acute lymphoblastic leukaemia (ALL), which might contribute to the wide variation in the reported incidence of myeloid antigen expressing ALL in childhood. METHODS--Stored leukaemic cells from 10 children with previously defined myeloid positive ALL were examined. A range of commercially available anti-CD13 and anti-CD33 monoclonal antibodies, directly conjugated with phycoerythrin or fluorescein isothyocyanate, or both, was used. Positively reacting cells were detected by flow cytometry. RESULTS--There was a noticeable discordance between the different commercial sources of antibody and between the two fluorochromes in their ability to detect myeloid antigens, as well as variation in the intensity of staining. For CD13, one antibody reacted with eight cases and another with only four. Similarly, CD33 was detected in all 10 cases by one antibody and in only three by another. CONCLUSIONS--The lack of any consistent pattern of results suggests that various commercial antibodies against the same CD antigen might recognise different epitopes and that the number of molecules per cell might vary from case to case. These observations partly explain the variation in reported incidence and the failure to establish the clinical importance of myeloid positivity, and they highlight the importance of standardisation in multicentre studies in which immunophenotypic data are collected.