Most breast cancer patients treated with trastuzumab (which has been the most effective therapy to date in HER-2+ cases) will not exhibit response or will soon develop treatment resistance. Patient selection has been focused on determination of HER-2+ status alone. It is uncertain whether HER-2 co-expression combined with other ERBB family receptors defines specific breast tumor subsets, but in vitro studies have demonstrated that heterodimerization of the ERBB family can promote malignant phenotypes. In this dissertation, I test the hypothesis that the expression of a combination of biomarkers within the ERBB family pathway is associated with breast cancer patient prognosis and response to therapy as well as specific programs of expression of downstream pathway mediators. I have documented several findings: (1) standardized assay conditions for assessment of biomarker expression by automated quantitative analysis (AQUA), (2) the prognostic effect of the ERBB family network in primary breast cancer, (3) the negative predictive effect of EGFR on tamoxifen treatment for ER+ premenopausal breast cancer, and (4) an AQUA-based biomarker assay for HER-2 that improves prediction of trastuzumab response in metastatic disease. These results were the first to demonstrate predictive advantage of a quantitative protein detection technique over conventional pharmacodiagnostic assays.