The National Institute of Neurological Disorders and Stroke (NINDS) convened experts in traumatic brain injury (TBI) research, policy, clinical practice and people with lived experience to propose a system of injury classification less susceptible to misinterpretation and misrepresentation inherent in the current use of "mild", "moderate" and "severe". One of six working groups addressed Retrospective Classification of TBI. The Working Group consisted of 14 experts in brain injury research representing a breadth of professional disciplines. Initial conclusions based on expert opinion were vetted and revised based on public input at the January 2024 NINDS TBI Classification and Nomenclature Workshop. The Working Group examined five types of methodologies for identifying past TBIs (self/proxy-report, medical record extraction, imaging, fluid-based biomarkers, and performance-based tests). They concluded that self/proxy-report is essential for clinical, research and surveillance applications and that clinicians and researchers should employ elicitation protocols that have been studied and found valid. Medical record extraction was also identified as an invaluable tool for identification of past history of medically attended TBIs; however, there is a need to standardize the case definition employed and procedures used. The use of imaging methods, fluid-based biomarkers, and performance-based assessments in isolation lacked sufficient evidence of both sensitivity and specificity in detecting past histories of TBI to be recommended for this use at this time. The Working Group also evaluated identification of repetitive head impacts (RHI), finding no evidence of a common definition of RHI, a requisite initial step for the development and validation of standardized instruments.