Background: ERBB2mutations () occur in 1-4% of non-small cell lung cancers (NSCLC). Trastuzumab-deruxtecan is FDA-approved for the treatment of metastatic ERBB2-mutant NSCLC based mostly on data from NSCLCs harboring tyrosine kinase domain (TKD) ERBB2mut. The genomic and clinical characteristics of NSCLC with ERBB2 mut beyond the TKD remain largely unexplored. Methods: Patients (pts) with NSCLCs with oncogenic ERBB2 mut and genomic tumor profiling data from the AACR Project GENIE database were included. The cohort was divided into pts with tumors harboring TKD, extracellular (ECD), or transmembrane (TMD) and juxtamembrane domain (JMD) ERBB2mut. Clinico-genomic characteristics were compared between the groups. The Kaplan-Meier method was used to estimate progression-free survival (PFS) with first-line systemic therapy, and log-rank tests were used for comparisons. Results: Of 483 pts with ERBB2-mutant NSCLC, 64% were female, 81% were White, and 13% were Asian. TKD, ECD, and TMD/JMD ERBB2 mut were identified in 362 (75%), 88 (18%), and 33 (7%) cases, respectively. Pts with ECD vs. TKD ERBB2mut tumors were more likely to be males (47% vs. 33%, p=0.03) and have tumors with squamous histology (7% vs. 1%, p<0.0001). Among pts with (n=39), those with ECD vs. TKD ERBB2 mut tumors were more likely to have a history of smoking (100% vs. 42%, p = 0.02). The most common TKD, ECD, and TMD/JMD ERBB2 mut were in-frame ex20ins Y772_A775dup (n=224/362, 62%), missense S310F/Y (n=45/88, 51%), and missense V659E/D (n=10/33, 30%) mut, respectively. ECD vs. TKD ERBB2-mutant tumors harbored a distinct genomic phenotype with significantly higher median tumor mutational burden (11.1 vs. 5.2 mut/Mb, p<0.003), median fraction of genome altered (0.2 vs. 0.1, p = 0.02), and rate of co-mut in EGFR (32% vs. 3.6%, p<0.0001), KRAS (17% vs. 2.5%, p<0.0001), STK11 (23% vs. 4.3%, p<0.0001), KEAP1 (23% vs 5%, p<0.0001) and SMARCA4 (19% vs 4.5%, p<0.0001). Of 343 cases with ERBB2 copy number profiling, 32 (9.3%) harbored ERBB2 amplification. ERBB2-mutant NSCLCs with vs. without ERBB2amplifications had a higher frequency of TP53 co-mut (84% vs. 48%, p <0.001). Of 39 pts with available clinical data, most had ECD (n=6) ERBB2 mut tumors and received first-line platinum-based chemotherapy (n=31). The presence of a co-mut in STK11 (HR 3.5, 0.4–27.4, p=0.03) and KEAP1 (HR 3.9, 0.7–20.6, p=0.002) conferred shorter PFS on first-line systemic therapy. Pts with ECD vs. TKD ERBB2 mut tumors had shorter PFS (HR 2.4, 95% CI 0.7–8.2, p=0.04). Conclusions: NSCLC tumors harboring ECD ERBB2 mut are associated with a distinct clinical and molecular phenotype and inferior outcomes with chemotherapy compared to pts with tumors harboring TKD mut. These findings enhance our understanding of disease heterogeneity in ERBB2-mutant NSCLC and may aid in optimizing treatment strategies for this unique population.