The promising effect of antibody–drug conjugates on breast cancer with low expression of HER2 (HER2-low) raises many questions regarding the optimal selection of patients for this treatment. A key question is whether HER2 immunohistochemistry, an assay optimised to detect HER2 amplification, is reliable enough to assess HER2 protein levels to select patients with HER2-low breast cancer in daily pathology practices worldwide. Moreover, whether this assessment can be performed with sufficient reproducibility between pathologists in daily practices is debatable. Herein, we address the historical track record of the CAP-ASCO HER2 Guidelines, the reported limited reproducibility by pathologists of HER2 immunohistochemistry in the non-amplified cases, and the performance variation of different antibodies. Based on this summary, we propose solutions to improve the robustness to enable reliable identification of patients with HER2-low breast cancer.