Abstract: Objective To explore the clinical and biological factors influencing the therapeutic response to pembrolizumab in patients with advanced non-small cell lung cancer（NSCLC）, and to establish a nomogram prediction model to assist individualized treatment decision-making. Methods A multicenter retrospective cohort design was adopted. A total of 247 patients with stage ⅢB-Ⅳ NSCLC admitted to The First People's Hospital of Changzhou, The First Affiliated Hospital of Soochow University and The Second Affiliated Hospital of Soochow University from January 2020 to December 2024 were enrolled as the development cohort. Demographic characteristics, tumor molecular biomarkers programmed death-ligand 1（PD-L1） and tumor mutational burden（TMB）,clinicopathological parameters, and inflammatory indicators neutrophil-to-lymphocyte ratio（NLR）, platelet-to-lymphocyte ratio（PLR）,and lymphocyte-to-monocyte ratio（LMR） were collected. Univariate and multivariate logistic regression were performed to screen factors associated with treatment efficacy and construct the nomogram model. The model performance was evaluated using receiver operating characteristic（ROC） curve, Bootstrap method and calibration curve. Meanwhile, 112 independently and prospectively recruited patients from The First People's Hospital of Changzhou between January 2025 and October 2025 were enrolled as an external validation cohort, and the model performance was verified using ROC curve, calibration curve and Bootstrap method. Results Univariate analysis revealed that ≥ 3 metastatic sites, negative PD-L1 expression, TMB < 10 mut·Mb^-1, monotherapy, NLR ≥ 5, PLR ≥150 and LMR ≤ 1.5 were significantly associated with poor efficacy（P< 0.05）. Multivariate analysis showed that ≥ 3 metastatic sites,negative PD-L1 expression, TMB <10 mut·Mb^-1, monotherapy, NLR ≥ 5, PLR ≥ 150, and LMR ≤ 1.5 were independent risk factors for poor efficacy of pembrolizumab（P< 0.05）. For the nomogram model, the area under the curve（AUC） for the ROC curve was 0.771（95% CI: 0.713-0.830）, and the concordance index（C-index） was 0.753（95% CI: 0.698-0.808）. The calibration curve showed good consistency between the predicted probabilities and actual observations（Hosmer-Lemeshow goodness-of-fit test: χ² = 7.820, P= 0.452）.In the external validation, the model achieved an AUC of 0.744（95% CI: 0.653-0.835）. The calibration curve had a slope of 0.921（95% CI: 0.852-0.990）, and the Hosmer-Lemeshow goodness-of-fit test yielded χ² = 5.210, P= 0.767, indicating good consistency between predicted probabilities and actual incidence rates. Conclusion The number of metastatic sites, PD-L1 expression status, TMB level, treatment regimen and inflammatory indicators are key predictors of pembrolizumab efficacy. The nomogram model constructed in this study has moderate-to-good predictive performance, and can provide a reference for clinical screening of patients who may benefit from pembrolizumab and optimizing treatment strategies.