Aim: To identify risk factors for self-reported cognitive impairment in radically treated stage III non-small cell lung cancer (NSCLC). Methods: Cognitive functioning was assessed using the EORTC-QLQ-C30 at seven pre-specified time points in the phase III NVALT-11 trial (observation versus prophylactic cranial irradiation [PCI] in stage III NSCLC treated with chemo-radiotherapy +/- surgery). Cognition was analyzed as binary (impairment or not) and continuous outcome, respectively, using generalized estimating equation (GEE) before and after multiple imputation. A score < 75 was defined as cognitive impairment. A mean difference by < 10, 10-<20, >= 20 points was regarded as of no, moderate, and large clinical effect, respectively. We categorized the cognitive impairment into four types based on changes over time: sustained, rever-sible, recurring, and alternating. Results: In the no-PCI arm, 43/84 [51.2%] reported cognitive impairment at least once, of which 31.4% were sustained, 25.7% reversible, 28.6% recurring, and 14.3% alternating. Results were similar in the PCI arm. Cognitive functioning at baseline was comparable in two arms and a score < 75 was a significant risk factor with large effect for subsequent cognitive impairment (no-PCI: p = -23.30, p < 0.001; PCI arm: p = -22.34, p < 0.001; All: p = -23.47, p < 0.001). Younger age (<60y), squamous histology, and PCI were risk factors without clinical relevance (p >-10, p < 0.05). Cognitive functioning declined over time (p = -0.26, p = 0.001) except for patients with cognitive impairment at baseline (p = 0.141, p = 0.33). Conclusion: Cognitive impairment is dynamic over time with four types. Baseline cognitive impairment (score < 75) is the most important risk factor for subsequent cognitive impairment in stage III NSCLC. Note: This work has been partly reported as an oral presentation at the ESTRO 2021 meeting (OC-0176). (c) 2023 Elsevier B.V. All rights reserved. Radiotherapy and Oncology xxx (2023) xxx-xxx