Breast cancer is a major health problem worldwide. In ~15% of breast cancers, the epidermal growth factor receptor HER2, a transmembrane protein, is overexpressed. This HER2 overexpression is associated with an aggressive form of the disease and a poor clinical prognosis. The extracellular domain (ECD) of HER2 is released into the blood by a proteolytic mechanism known as "ECD shedding". This proteolytic shedding leaves a constitutively active truncated receptor in the membrane that is 10-100-fold more oncogenic than the full-length receptor and promotes the growth and survival of cancer cells. Shedding of the HER2 ECD is increased during metastasis: whereas 15% of primary breast cancer patients have elevated levels of serum HER2 ECD (sHER2 ECD), the levels reach 45% in patients with metastatic disease. Thus, sHER2 ECD has been proposed as a promising biomarker for cancer recurrence and for monitoring the disease status of patients overexpressing HER2. Nevertheless, in 2016, the American Society of Clinical Oncology advises clinicians not to use soluble HER2 levels to guide their choice of adjuvant therapy for patients with HER2-positive breast cancer, because the evidence was considered not strong enough. Currently, biomarkers such as carcinoembryonic antigen and cancer antigen 15-3 are widely used to monitor metastatic breast cancer disease even if the level of evidence of clinical impact of this monitoring is poor. In this article, we review the evidence that sHER2 ECD might be used in some situations as a biomarker for breast cancer. Although this serum biomarker will not replace the direct measurement of tumor HER2 status for diagnosis of early-stage tumors; it might be especially useful in metastatic disease for prognosis, as an indicator of cancer progression and of therapy response, particularly to anti-HER2 therapies. Owing to these data, sHER2 ECD should be considered as a promising biomarker to detect cancer recurrence and metastasis.