Targeting EGFR is a validated approach in the treatment of squamous-cell cancers (SCCs), although there are no established biomarkers for predicting response. We have identified a synonymous mutation in EGFR, c.2361G>A (encoding p.Gln787Gln), in two patients with head and neck SCC (HNSCC) who were exceptional responders to gefitinib, and we showed in patient-derived cultures that the A/A genotype was associated with greater sensitivity to tyrosine kinase inhibitors (TKIs) as compared to the G/A and G/G genotypes. Remarkably, single-copy G>A nucleotide editing in isogenic models conferred a 70-fold increase in sensitivity due to decreased stability of the EGFR-AS1 long noncoding RNA (lncRNA). In the appropriate context, sensitivity could be recapitulated through EGFR-AS1 knockdown in vitro and in vivo, whereas overexpression was sufficient to induce resistance to TKIs. Reduced EGFR-AS1 levels shifted splicing toward EGFR isoform D, leading to ligand-mediated pathway activation. In co-clinical trials involving patients and patient-derived xenograft (PDX) models, tumor shrinkage was most pronounced in the context of the A/A genotype for EGFR-Q787Q, low expression of EGFR-AS1 and high expression of EGFR isoform D. Our study reveals how a 'silent' mutation influences the levels of a lncRNA, resulting in noncanonical EGFR addiction, and delineates a new predictive biomarker suite for response to EGFR TKIs.