Background: Programmed death-ligand 1 (PD-L1) serves as a major predictive biomarker for immune checkpoint inhibitors in non-small cell lung cancer (NSCLC). However, the relationship between PD-L1 expression and genetic features of NSCLC remains unclear. The aim of this study was to explore the correlation between genetic profiles and PD-L1 expression in NSCLC. Methods: FFPE tumor and matched blood samples from 568 NSCLC (487 adenocarcinomas (ACAs) and 81 squamous carcinomas (SCAs)) patients were collected for NGS-based targeted panel sequencing. Genomic alterations including single nucleotide variations, short and long insertions/deletions, copy number variations and gene rearrangements were assessed. Tumoral PD-L1 expression was evaluated by immunohistochemical analysis (Dako 22C3 and 28-8). Results: The prevalence of PD-L1 expression was 9.9% with a ≥50% cutoff and 27.5% with a ≥1% cutoff. High PD-L1 expression (using a ≥1% cutoff) in tumor cells was significantly associated with mutations of MET (p < 0.001), RET (p = 0.036), ROS1 (p < 0.001), TP53 (p = 0.0013) and 11q13 amplification (p = 0.004), and was inversely correlated with EGFR mutations (p = 0.011). SOX2 and KLHL6 mutations were more frequent in SCAs (33.3% and 28.4%, respectively) than in ACAs (0.4% and 1%, respectively) and were adversely associated with PD-L1 expression in SCAs (p = 0.01, p = 0.004). Gene set 1 (GS1) mutations included EGFR, ALK, KRAS, BRAF, MET, RET, ROS1 and ERBB2 and gene set 2 (GS2) mutations included TP53, RB1, PTEN, APC and MYC. According to the mutation status of GS1 and GS2, 18 patients were classified as type I (GS1-, GS2-), 92 as type II (GS1-, GS2+), 202 as type III (GS1+, GS2-) and 256 as type IV (GS1+, GS2+). PD-L1 expression was higher in Type IV tumors than in type III (33.2% vs. 14.4%, respectively, p < 0.001). Conclusions: We highlighted the genomic heterogeneity of NSCLC according to the mutation status of different gene sets. Our results indicate that patients with GS1 and GS2 gene mutations might correlate with higher PD-L1 expression. Our results help to understand the relationship between genomic features and PD-L1 expression and may be a potential guide for immunotherapy.