Triple‐negative breast cancer (TNBC) represents 10% of all breast cancers and is a very heterogeneous disease. Globally, women with TNBC have a poor prognosis, and the development of effective targeted therapies remains a real challenge. Patient‐derived xenografts (PDX) are clinically relevant models that have emerged as important tools for the analysis of drug activity and predictive biomarker discovery. The purpose of this work was to analyze the molecular heterogeneity of a large panel of TNBC PDX (n = 61) in order to test targeted therapies and identify biomarkers of response. At the gene expression level, TNBC PDX represent all of the various TNBC subtypes identified by the Lehmann classification except for immunomodulatory subtype, which is underrepresented in PDX. NGS and copy number data showed a similar diversity of significantly mutated gene and somatic copy number alteration in PDX and the Cancer Genome Atlas TNBC patients. The genes most commonly altered were TP53 and oncogenes and tumor suppressors of the PI3K/AKT/mTOR and MAPK pathways. PDX showed similar morphology and immunohistochemistry markers to those of the original tumors. Efficacy experiments with PI3K and MAPK inhibitor monotherapy or combination therapy showed an antitumor activity in PDX carrying genomic mutations of PIK3CA and NRAS genes. TNBC PDX reproduce the molecular heterogeneity of TNBC patients. This large collection of PDX is a clinically relevant platform for drug testing, biomarker discovery and translational research. What's new? Triple‐negative breast cancer (TNBC) is a highly heterogeneous disease in terms of molecular profile, histological features, clinical behavior, and drug response. Several clinical trials have been conducted for targeted therapies, but only in unselected TNBC patients, with disappointing results. This study shows that patient‐derived xenografts (PDX) reproduce the molecular heterogeneity of TNBC. The presented genomic analysis identifies several interesting targetable drivers, particularly in the PI3K and MAPK pathways. PDX models provide a unique opportunity to test various treatments on individual tumors: already, two specific inhibitors (dual PI3K/mTOR and MEK inhibitor) and their combination are providing encouraging results.