Background: Historically, racial and ethnic minorities, women, and older patients have been underrepresented in RCTs. Hence, therapeutics are approved based on data not representative of the broader population, leaving important gaps in knowledge of safety and effectiveness. RWE may help address these gaps and inform regulatory, clinical, payer, and manufacturer decision-making. Racial, sex at birth, and age distributions were compared between RWE studies and a corresponding RCT in the aNSCLC setting. Additionally, RCT data were compared to SEER aNSCLC data. Methods: Demographics of adults with aNSCLC in the KEYNOTE-407 RCT were compared with those of adults in two physician-abstracted medical chart review US RWE studies conducted in 2019 or 2020 or with the most recently available US SEER data (2017-2019). Statistical comparisons were performed using t-tests or chi square tests. Results: We identified 10,089 patients (RCT: n = 559, 5.5%; RWE: n = 783, 7.8%; SEER: n = 8,747, 86.7%) (Table). Significant differences in racial distribution were observed between the RCT and RWE study populations, with Black/African American patients comprising 1.3% of RCT patients and 24.0% of the RWE study population. Non-Hispanic Black patients made up 12.8% of SEER patients. The proportion of female patients was lower in the RCT (18.6%) compared with RWE (40.5%) and SEER (36.7%) data. Age at 1L initiation was significantly lower for RCT patients vs. age at metastatic diagnosis in SEER (mean: 64.7 vs. 70.0 years). However, no statistically significant difference in age at 1L initiation was observed between RCT and RWE data (64.9 vs. 64.7 years). Conclusions: In aNSCLC, significant differences in sex at birth and race were observed between RCT and RWE study populations. Additionally, significant differences in sex at birth and age were observed between RCT and SEER data. Given that regulatory, therapeutic development, clinical, payer, and decision-making is largely driven by data from RCTs, future research can leverage RWE to improve representativeness and generalizability.