Pulmonary large-cell carcinoma-a diagnostically and clinically controversial entity-is defined as a non-small-cell carcinoma lacking morphologic differentiation of either adenocarcinoma or squamous cell carcinoma, but suspected to represent an end stage of poor differentiation of these tumor types. Given the recent advances in immunohistochemistry to distinguish adenocarcinoma and squamous cell carcinoma, and the recent insights that several therapeutically relevant genetic alterations are distributed differentially in these tumors, we hypothesized that immunophenotyping may stratify large-cell carcinomas into subsets with distinct profiles of targetable driver mutations. We therefore analyzed 102 large-cell carcinomas by immunohistochemistry for TTF-1 and ΔNp63/p40 as classifiers for adenocarcinoma and squamous cell carcinoma, respectively, and correlated the resulting subtypes with nine therapeutically relevant genetic alterations characteristic of adenocarcinoma (EGFR, KRAS, BRAF, MAP2K1/MEK1, NRAS, ERBB2/HER2 mutations and ALK rearrangements) or more common in squamous cell carcinoma (PIK3CA and AKT1 mutations). The immunomarkers classified large-cell carcinomas as variants of adenocarcinoma (n=62; 60%), squamous cell carcinoma (n=20; 20%) or marker-null (n=20; 20%). Genetic alterations were found in 38 cases (37%), including EGFR (n=1), KRAS (n=30), BRAF (n=2), MAP2K1 (n=1), ALK (n=3) and PIK3CA (n=1). All molecular alterations characteristic of adenocarcinoma occurred in tumors with immunoprofiles of adenocarcinoma or marker-null, but not in tumors with squamous immunoprofiles (combined mutation rate 50% vs 30% vs 0%, respectively; P<0.001), whereas the sole PIK3CA mutation occurred in a tumor with squamous profile (5%). Furthermore, marker-null large-cell carcinomas were associated with significantly inferior disease-free (P<0.001) and overall (P=0.001) survival. In conclusion, the majority (80%) of large-cell carcinomas can be classified by immunomarkers as variants of adenocarcinoma or squamous cell carcinoma, which stratifies these tumors into subsets with a distinct distribution of driver mutations and distinct prognoses. These findings have practical implications for diagnosis, predictive molecular testing and therapy selection.