Objective: Retrospective concussion studies rely on self-report, which is often reported as a limitation, particularly when injuries occurred years or decades earlier. Since little is known about concussion reporting reliability and whether cognitive impairment may impact such reports, we examined the reliability of self-reported remote concussion history within a sample of retired professional football players with and without cognitive impairment. Method: Nineteen retired NFL players age 47–77 years (M = 62.9) were recruited from a longitudinal study, including eight with cognitive impairment. Cognitive status (normal vs. impaired) was determined by consensus of a behavioral neurologist and neuropsychologist using interviews and neuropsychological data. Self-reported history of lifetime concussions was obtained at baseline and again at 12-25 months (M = 19.5). The behavioral neurologist graded concussions [loss of consciousness (LOC) = grade 3]. Correlational analyses assessed reliability of reports between time points. Results: Reported history of concussion with LOC and total number of grade 3 concussions were significantly correlated between time points for the total sample (rs = .89 and .76). Similar results were found in the subgroup of eight cognitively impaired players (rs = 1.0 and .79). In contrast, the correlation between total number of concussions (with or without LOC) was not significant between time points in either the total sample (r = .37) or the cognitively impaired subgroup (r = .22). Conclusion: Although further investigation is needed, self-reported concussion history with LOC appears to be reliable among retired NFL players with or without cognitive impairment. Retrospective estimates of total number of concussions with or without LOC, on the other hand, appears less reliable regardless of cognitive status.