Abstract Introduction The status of the gene encoding human EGF-like receptor 2 (HER2) is an important prognostic and predictive marker in breast cancer. Only breast cancers with HER2 amplification respond to the targeted therapy with trastuzumab. It is controversial to what degree the primary tumour is representative of distant metastases in terms of HER2 status. Discrepancies in HER2 status between primary tumours and distant metastases have been described, but their reasons remain unclear. Here, we compared HER2 status on cytological specimens of distant metastases with the result from the primary carcinomas, and explored the prevalence of and the reasons for discrepant results. Methods HER2 status was determined by fluorescence in situ hybridisation. HER2 gene amplification was defined as a HER2/chromosome 17 signal ratio of 2 or more. HER2 results from cytological specimens of matched distant metastases were compared with the results from the corresponding primary tumours (n = 105 patients). In addition, lymph node metastases were analysed in 31 of these patients. Results HER2 amplification was found in 20% of distant metastases. HER2 status was discordant between the primary tumour and distant metastasis in 7.6% of the 105 patients. Re-evaluation revealed that in five patients (4.7%), discrepancies were due to interpretational difficulties. In two of these patients, focal amplification had initially been overlooked as a result of heterogeneity in the primary tumours or in the metastases, respectively. A further three patients had borderline amplification with a ratio close to 2. Discrepancy remained unexplained in three patients (2.9%). Conclusion HER2 gene status remains highly conserved as breast cancers metastasise. However, discrepant results do occur because of interpretational difficulties and heterogeneity of HER2 amplification. Cytological specimens from distant metastases are well suited for HER2 fluorescence in situ hybridisation analysis.