Gene expression analysis on messenger RNA (mRNA) purified from formalin-fixed, paraffin-embedded tissue is increasingly used for research purposes. Tissue heterogeneity may question specificity and interpretation of results from mRNA isolated from a whole slide section, and thresholds for minimal tumor content in the paraffin block or macrodissection are used to avoid contamination from non-neoplastic tissue. The aim was to test if mRNA from tissue surrounding breast cancer affected quantification of estrogen receptor α (ESR1), progesterone receptor (PGR) and human epidermal growth factor receptor 2 (ERBB2), by comparing gene expression from whole slide and tumor-enriched sections, and correlating gene expression from whole slide sections with corresponding immunohistochemistry. Gene expression, based on mRNA extracted from a training set (36 paraffin blocks) and two validation sets (133 + 1,083 blocks), were determined by quantitative reverse transcription polymerase chain reaction for all samples, as well as by microarray for 133 validation samples. In the training set, agreement between high vs. low mRNA expression from whole slide and tumor-enriched sections was absolute for ESR1 and ERBB2, and 83 % for PGR. Overall agreements, when comparing mRNA expression to immunohistochemistry, were 100 % (ERBB2), 89 % (ESR1) and 83 % (PGR), which was confirmed in the validation sets. Percentage of tumor in the sections did not influence the results. In conclusion, reliable quantification of ESR1, PGR and ERBB2 mRNA expression can be obtained from a whole slide section, and correlates well with immunohistochemistry. Prior removal of surrounding tissue was found to be unnecessary even with minimal tumor content in the section.