Background: Nivolumab, a PD-1 inhibitor, has demonstrated meaningful clinical responses in non-small cell lung cancer (NSCLC) patients. However, long-term survival data remain limited, necessitating predictive modeling to assess its sustained clinical benefit. Methods: A systematic review was conducted across PubMed, Embase, and the Cochrane Library (search cutoff: March 24, 2025), from which 27 clinical trials were selected. Time-to-event (TTE) data were extracted from Kaplan–Meier (KM) curves for overall survival (OS), with covariates derived from reported demographic. Parametric TTE models were developed using Monolix 2024R1 (Lixoft, Antony, France) and validated by visual predictive checks (VPCs). Results: IPD were reconstructed from 2764 and 3608 patients for first-line and second-line therapy. Across all evaluated models, the log-normal distribution demonstrated the best fit according to VPCs and objective function value comparisons. Separate models developed for each therapy line provided a better fit for both the OS curve shape and the long-term tail compared with the pooled model. Conclusion: Whether the first-line or second-line therapy is a key determinant of survival in nivolumab-treated NSCLC patients. Further analyses involving the development of separate models for each line of therapy may enable a more detailed evaluation of covariate effects.