BCR-ABL fluorescence in situ hybridization has a useful role to play in experimental and clinical investigations of chronic myeloid leukaemia. However, the interpretation of results is complicated by variability in the false positive rate (FPR) and false negative rate (FNR). We therefore examined the effects on FNR and FPR of three factors, namely, the criteria used for defining a fusion signal, nucleus size, and the genomic position of the ABL breakpoint. We established two different criteria for BCR-ABL positivity: by criterion A cells were scored as positive when BCR and ABL signals were overlapping or touching and by criterion B cells were positive if they satisfied criterion A or if the signals were separated by up to one signal diameter. We measured nucleus size and Philadelphia (Ph) positivity in 573 cells from normal persons and 787 cells from the Ph+ SD-1 cell line and related results to FNRs and FPRs. We also assessed the FNR in Ph+ CFU-GM colonies from five patients with different ABL breakpoints. We showed that each of these factors influenced the FNR and FPR. The less strict criterion (B) for Ph positivity increased the FPR but reduced the FNR, the FPR increased as the nucleus size decreased, and the FNR was greatest in CML cells with a 5' ABL breakpoint. We conclude that these factors should be considered when evaluating the results of FISH studies to detect the BCR-ABL fusion gene and that analogous factors may influence results of FISH studies directed at other fusion genes.