Background: Neutrophil to lymphocyte ratio (NLR) predicts overall survival (OS) in cancer patients (pts) treated with immune checkpoint inhibitors (ICI). Historically, most studies included heterogenous populations and had limited data beyond baseline values. We further studied its prognostic value in a more homogenous cohort of only metastatic Non–Small Cell lung cancer (NSCLC) pts who have received first-line pembrolizumab-based therapy over the first 3 cycles. Methods: This was a cohort study with 282 pts from 2017–2021. NLR = absolute neutrophil/lymphocyte count from complete blood count (CBC). Baseline CBCs were collected within 7 days of initiating ICI. Cycle 2 and 3 CBCs were collected with corresponding treatments. Due to its non-normal distribution, NLR was natural log-transformed before analyses were performed. Cox proportional hazard model was used to examine the association between baseline NLR and OS with p<0.1. ΔNLR (difference) was calculated between each cycle and baseline. Cox model was used to test the association of time-dependent ΔNLR with OS using counting process form, controlling for baseline NLR. Nature cubic splines were used to explore the non-linear associations between baseline NLR or ΔNLR with OS. One patient was excluded from cubic splines for having very low NLR < 1. All calculations were performed using SAS V9.4. Results: In our study, median age was 63.7 years with interquartile range of 56.5–72.0. 120 (42.6%) received pembrolizumab monotherapy, and 162 (57.4%) had pembrolizumab with chemotherapy. 125 (44.3%) were females. 197 (69.9%) had adenocarcinoma, and 85 (30.1%) had other histology. 115 (40.8%) pts had PDL1 ≥50%, 68 (24.1%) pts had 1–49%, 93 (33.0%) pts had no PDL1 expression, and 6 (2.1%) pts with unknown PDL1. On multivariable analysis, one unit increased in log-baseline NLR was associated with 24.0% higher hazard of death (HR=1.24, 95% CI=(1.03, 1.50), p=0.0261) after controlling for age, ECOG status, histology, immunotherapy, PDL1, sex, and race. For one unit increased in ΔNLR, the hazard of death increased by 69.2% (HR=1.69, 95% CI=(1.29, 2.22), p=0.0001). Adjusted cubic splines (Figure 1a and 1b) demonstrated an overall linear increase in mortality risk with increasing log-baseline NLR or ΔNLR. Conclusion: High NLR at baseline and increasing over time was associated with reduced overall survival in metastatic NSCLC patients who have received pembrolizumab-based treatment as the first-line therapy.