The HER2 extracellular domain shed in blood (HER2 ) is reported to rise and fall in parallel with HER2+ breast cancer behavior. In this study, we evaluated the clinical relevance of plasma HER2 values in patients with metastatic breast cancer treated in the SAKK22/99 trial comparing trastuzumab monotherapy followed by trastuzumab-chemotherapy combination at progression versus upfront combination therapy. Quantitative assessment of plasma HER2 was performed in 133 patients at baseline; after 2-24 h; at 3 weeks; at first response evaluation (8-9 weeks); and at tumor progression. Associations with tumor characteristics, disease course and trial treatment were evaluated. Baseline HER2 levels were stable within 24 h after the first trastuzumab injection. These plasma values correlated positively with the HER2 gene ratio (r  = 0.39, P < 0.001) and HER2 protein expression levels (r  = 0.36, P < 0.001) but not with ER/PR status of the primary tumor. HER2 baseline levels were positively associated with the presence of visceral disease (P = 0.05) and poor patients' outcome (Cox-regression: P = 0.009). Patients with high baseline levels (> 35 ng/ml) had the worst overall survival (P = 0.03) if treated with upfront combination therapy. Conversely, patients with low HER2 baseline values (< 15 ng/ml) had longer time to progression on combined trastuzumab-chemotherapy when first treated with trastuzumab monotherapy (P = 0.02). Monitoring HER2 levels during the course of the trial revealed significant time (P = 0.001) and time-treatment arm interactions (P = 0.0007). Under upfront trastuzumab alone, the HER2 levels remained stable until just before disease progression. In patients responding to combination treatment HER2 levels decreased to > 20%. Plasma HER2 levels in patients with metastatic breast cancer reflect HER2 disease status. This robust biomarker might help identifying patients without visceral disease profiting from a sequential treatment's modality. Monitoring HER2 levels during trastuzumab monotherapy could help defining the optimal time to introduce chemotherapy. Registration Number by ClinicalTrials.gov: NCT00004935, Trial number: SAKK22/99. Registered on 27 January 2003.