The intra-patient variations of some clinically relevant quantifiable features, between axillary lymph node metastases were evaluated in 44 breast cancer patients. In all lymph node metastases detected (range 2-33 per patient), the mitotic figures were counted, the volume percentage epithelium was assessed and the mean nuclear area was measured. The intra-patient variation for each quantifiable feature was expressed by the coefficient of variation (CV). Since the measurement techniques used introduce a certain, well known variation themselves because of sampling and measurement errors, the CVs found had to be greater than methodological tolerance limits (established in previous studies) to be interpreted as indicating biological variation. The CVs exceeded the methodological tolerance limits in 86% of the cases for the mitotic count, in 48% of the cases for the volume percentage epithelium, and in 47% of the cases for the mean nuclear area. This indicated that in these cases, the variation found in the quantifiable features could not be explained by sampling or measurement errors and should be regarded as real biological variation. Furthermore, the variation in the quantifiable features studied showed a significant positive correlation with the number of lymph node metastases. Thus, there may be considerable intra-patient variation in quantifiable features between axillary lymph node metastases in breast cancer. This may indicate that these lymph node metastases originate independently from different clones within the primary tumour, that they are independently formed in different stages of tumour development, or that they, as an expression of intrinsic tumour heterogeneity, may develop in different directions from the start.