Background/AimThe relationship between diabetic retinopathy (DR) and cognitive impairment (CI) is unclear due to equivocal findings from cross-sectional studies and a lack of long-term data. In this population-based cohort study, we investigated the longitudinal association between the severity of DR and the incidence of CI.Methods682 participants with diabetes, gradable retinal photographs and no CI at baseline 2004–2011) and complete relevant data at follow-up 2010–2016 from the Singapore Epidemiology of Eye Disease Study were included. CI was assessed using the validated Abbreviated Mental Test (AMT), defined as scores of ≤6 and ≤8 for those with 0–6 and >6 years of formal education, respectively. Six-year incident CI was defined as having no CI at baseline but present at the follow-up visit.ResultsOf the 682 included participants, 483 (70.8%) had no DR and 199 (29.2%) had any DR. Of those with DR, 142 (20.8%) had minimal/mild DR and 57 (8.4%) had moderate or worse DR at baseline. At the follow-up visit, 40 (5.9%) participants had incident CI based on AMT. In multivariate analysis compared with participants without DR, those with any DR had more than twofold increased odds of incident CI (OR (95% CI): 2.32 (1.07 to 5.03)). Participants with moderate or worse DR had threefold increased odds of developing CI (3.41 (1.06 to 11.00)), compared with those with no DR.ConclusionsDR, particularly at the more severe stages, is associated with increased risk of developing CI, independent of vision and other risk factors.