We examined the relationship between retinal nerve fiber layer (RNFL) thickness and cognitive function in a population of older British adults.Participants of the European Prospective Investigation of Cancer (EPIC) Norfolk cohort study underwent ophthalmic and cognitive assessment. Measurements of RNFL thickness were made using the Heidelberg Retina Tomograph (HRT). Cognitive testing included a short form of the Mini-Mental State Examination (SF-MMSE), an animal naming task, a letter cancellation task, the Hopkins Verbal Learning Test (HVLT), the National Adult Reading Test (NART), and the Paired Associates Learning Test. Multivariable linear regression models were used to assess associations of RNFL thickness with cognitive test scores, adjusted for age, sex, education level, social class, visual acuity, axial length, and history of cataract surgery.Data were available from 5563 participants with a mean age of 67 years. A thicker HRT-derived RNFL thickness was associated with better scores for the SF-MMSE (0.06; 95% confidence interval [CI], [0.02, 0.10], P = 0.005), HVLT (0.16, 95% CI [0.03, 0.29]; P = 0.014), and NART (-0.24, 95% CI [-0.46, -0.02], P = 0.035). The associations of RNFL thickness with SF-MMSE and HVLT remained significant following further adjustment for NART.We found a significant association between HRT-derived RNFL thickness and scores from cognitive tests assessing global function, recognition, learning, episodic memory, and premorbid intelligence. However, the associations were weak and not currently of predictive value. Further research is required to confirm and clarify the nature of these associations, and identify biological mechanisms.