Triple negative breast cancer (TNBC) refers to tumors that do not express clinically significant levels of estrogen and progesterone receptors, and lack membrane overexpression or gene amplification of ErbB-2/HER2, a receptor tyrosine kinase. Transcriptome and proteome heterogeneity of TNBC poses a major challenge to precision medicine. Clinical biomarkers and targeted therapies for this disease remain elusive, so chemotherapy has been the standard of care for early and metastatic TNBC. Our present findings placed ErbB-2 in an unanticipated scenario: the nucleus of TNBC (NErbB-2). Our study on ErbB-2 alternative splicing events, using a PCR-sequencing approach combined with an RNA interference strategy, revealed that TNBC cells express either the canonical (wild-type) ErbB-2, encoded by transcript variant 1, or the non-canonical ErbB-2 isoform c, encoded by alternative variant 3 (RefSeq), or both. These ErbB-2 isoforms function in the nucleus as transcription factors. Evicting both from the nucleus or silencing isoform c only, blocks TN cell and tumor growth. This reveals not only NErbB-2 canonical and alternative isoforms role as targets of therapy in TNBC, but also isoform c dominant oncogenic potential. Furthermore, we validated our findings in the clinic and observed that NErbB-2 correlates with poor prognosis in primary TN tumors, disclosing NErbB-2 as a novel biomarker for TNBC. Our discoveries challenge the present scenario of drug development for personalized BC medicine that focuses on wild-type RefSeq proteins, which conserve the canonical domains and are located in their classical cellular compartments.