Background: Immune checkpoint inhibitors (ICIs) targeting PD-1 or PD-L1 improved the survival of non-small cell lung cancer (NSCLC) patients with PD-L1 expression ≥50% and without alterations in EGFR, ALK, ROS1, RET. However, markers able to predict the efficacy of ICIs, in combination with PD-L1 expression are still lacking. Our aim in this hypothesis-generating pilot study was to evaluate whether the KRAS G12C variant may predict the efficacy of ICIs in advanced NSCLC patients with PD-L1 ≥ 50%. Methods: Genomic DNA or tissue sections of 44 advanced ICI-treated NSCLC cases with PD-L1 ≥ 50% without EGFR, ALK, ROS1, RET alterations were tested using Next Generation Sequencing, Fluorescence in Situ Hybridization and immunohistochemistry. Statistical analyses were carried out fitting univariate and multivariate time to event models. Results: KRAS G12C mutant patients (N = 11/44) showed a significantly longer progression-free survival (PFS) at univariate and multivariate analyses (p = 0.03). The Kaplan–Meier plot of the PFS time-to-event supports that G12C positive patients have a longer time to progress. PFS improvement was not observed when any KRAS mutations were compared to wild-type cases. Conclusions: Given the limitations due to the small sample size and exploratory nature of this study, we tentatively conclude the KRAS G12C mutation should be considered in future trials as a predictive marker of prolonged response to first-line ICIs in NSCLC patients overexpressing PD-L1. This finding could be relevant as anti-KRAS G12C therapies enter the therapeutic landscape of NSCLC.