Integrated assessment of tumour genomics and immune biomarkers may improve prognostic stratification in non-small cell lung cancer (NSCLC). We examined the distribution and prognostic relevance of somatic pathogenic/likely pathogenic variants (PVs), somatic copy-number alterations (SCNAs), programmed death-ligand 1 (PD-L1) expression and tumour-infiltrating lymphocytes (TILs) in a multicentre cohort. This retrospective, multicentre study included 367 unselected patients with histologically confirmed NSCLC treated at 19 centres in Greece between 2000 and 2020. Targeted next-generation sequencing (NGS) was performed in 267 tumours; PD-L1 tumour proportion score (TPS) and haematoxylin and eosin (H&E)-based TIL density were evaluated in 298 and 352 tumours, respectively. Overall survival (OS) was assessed using Kaplan–Meier and multivariable Cox regression analyses. TP53 (53.6%) and KRAS (16.5%) were the most frequently detected genes with PVs; SCNAs were identified in 22.5% of sequenced tumours and were more frequent in squamous than non-squamous histology. In multivariable analysis (254 complete cases), TIL density ≥10% (hazard ratio [HR] 0.72, 95% confidence interval [CI] 0.52–0.99; p = 0.042) and ≥2 mutated genes (HR 1.44, 95% CI 1.08–1.92; p = 0.014) were independently associated with OS. PIK3CA-containing copy-number gains were significantly enriched in squamous tumours with low TIL density (p = 0.024), while FGFR1-containing copy-number gains showed a similar but non-significant trend (p = 0.088). In the present retrospective study, co-mutation burden and TIL density provided complementary prognostic information beyond clinicopathological variables in NSCLC. The association between selected copy-number gains and an immune-poor squamous phenotype warrants prospective validation.