The aim of this study was to investigate to what extent mitotic activity index assessments are influenced by tumour heterogeneity. Ten invasive breast carcinomas of varying size were completely embedded, yielding 2-7 paraffin blocks per tumour. From each block, three H&E stained skip sections were cut and, in all sections, the mitotic activity index (MAI) and the mitoses per volume (M/V-)index were assessed. Coefficients of variation were calculated for the three different sampling levels (tumours, blocks, sections). In addition we recorded in how many tumours threshold discrepancies (values on both sides of the prognostic thresholds) occurred. Nested analysis of variance showed that most of the total variance occurred between the tumours. Threshold discrepancies occurred in four (40%) and five tumours (50%) for mitotic activity index and the mitoses per volume index, respectively. However, when grouping all sections from the same tumour and subjectively selecting the one showing the highest proliferation, the section with the highest mitotic activity was selected in nine of the 10 cases. In the other single case a prognostically correct value was still obtained. Thus, there is a noteworthy intra-tumour heterogeneity in mitotic activity in invasive breast cancer. We therefore propose the need to take multiple blocks per tumour and carefully scan all sections for the highest proliferative area.