While results thus far demonstrate the clinical benefit of trastuzumab, some patients do not respond to this therapy. To identify a molecular predictor of trastuzumab benefit, we conducted whole-transcriptome analysis of primary HER2+ breast carcinomas obtained from patients treated with trastuzumab-containing therapies and correlated the molecular portrait with treatment benefit. The estimated association between gene expression and relapse-free survival allowed development of a trastuzumab risk model (TRAR), with ERBB2 and ESR1 expression as core elements, able to identify patients with high and low risk of relapse. Application of the TRAR model to 24 HER2+ core biopsies from patients treated with neo-adjuvant trastuzumab indicated that it is predictive of trastuzumab response. Examination of TRAR in available whole-transcriptome datasets indicated that this model stratifies patients according to response to trastuzumab-based neo-adjuvant treatment but not to chemotherapy alone. Pathway analysis revealed that TRAR-low tumors expressed genes of the immune response, with higher numbers of CD8-positive cells detected immunohistochemically compared to TRAR-high tumors. The TRAR model identifies tumors that benefit from trastuzumab-based treatment as those most enriched in CD8-positive immune infiltrating cells and with high ERBB2 and low ESR1 mRNA levels, indicating the requirement for both features in achieving trastuzumab response.