Background: Diabetic retinopathy (DR) is a chronic diabetes complication. People with type 2 diabetes mellitus (T2DM) have two times the risk for dementia, suggesting it is a new chronic diabetes complication. Objective: Evaluate the association of DR with cognitive performance in a T2DM population. Methods: Cross-sectional study with 400 T2DM adults from whom socio-demographic, clinical, laboratory data, screening test for depression symptoms [Patient Health Questionaire-9 (PHQ-9) ], Mini Mental State Examination (MMSE) , Semantic Verbal Fluency test, Trail Making test A and B, Word Memory test were collected and performed. All cognitive test scores were converted into Global Cognition z-Score [GCS (z) ]. The association between GCS (z) < 0 with DR was performed using backward multivariate binary logistic regression model adjusted forage ≥ 65 years, school years ≤ 6 years, DM duration ≥ 10 years, depression symptoms score > 9 at PHQ-9, arterial hypertension, physical activity, diabetic retinopathy, macular edema and cardiovascular disease. Results: After exclusions, the 251 eligible patients were 56.6% female, mean age 61.1 (±9.8) years, DM duration of 12.6 (±8.9) years and 7.6 (±4.2) years of school education. DR prevalence was 46.5%. Multivariate Binaric Logistic Regression Model showed an association between DR and GCS (z) < 0, with odds ratio (CI95%) of 2.50 (1.18-5.34) , adjusted for age, low education level, arterial hypertension and depression symptoms [OD and CI95%respectively: 5.46 (2.42-12.34) ; 12.19 (5.62-26.46) ; 2.55 (0.88-7.39) ; 3.53 (1.55-8.07) ]. Conclusion: In this T2DMpopulation, having DR increased the chance for worse cognitive performance even when adjusted for age, low education level, presence of arterial hypertension, and depression symptoms. Disclosure A. R. A. Faria: None.