No biomarker beyond HER2 itself, which suffers from a low positive predictive value, has demonstrated clinical utility in breast cancer, despite numerous attempts to improve treatment tailoring for the growing number of anti-HER2 targeted therapies. This prompted us to examine the body of evidence, using a systematic approach, to identify putative predictive biomarkers in HER2-positive breast cancer, and discuss the hitherto failure to address the needs of patients. In the future, it is hoped immune-based biomarkers will predict benefit from anti-HER2 treatments in the neoadjuvant and adjuvant settings. In advanced-stage disease, the quantification of tumour heterogeneity using molecular-imaging technology has generated informative data on the success or failure of the antibody-drug conjugate T-DM1. Treatment tailoring remains a high priority, in cost-constrained health-care systems, but such tailoring will require a dramatic shift in the way translational research is being conducted, with the establishment of large, easily accessible, and well-annotated databases of candidate predictive biomarkers. Single-centre biomarker research should become a thing of the past.