BACKGROUND: Individuals with type 2 diabetes have an increased risk of developing both vascular and Alzheimer's dementia. OBJECTIVE: This prospective cross-sectional study assessed the screening ability of the standard Montreal Cognitive Assessment (MoCA) score suggestive of mild cognitive impairment (<
26) in a European cohort of individuals ≥65 of age with type 2 diabetes. METHODS: Participants of RECOGNISED, a European prospective EU-funded cohort study, were screened using MoCA. In addition, a 13-item Neuropsychological Test Battery (NTB) with the Clinical Dementia Rating was undertaken to categorize participants as normocognitive (NC, n = 128) or mild cognitive impaired (MCI, n = 185). Receiver operating characteristic (ROC) analysis was used to evaluate the ability of MoCA cut-off scores to categorize patients as having MCI or not. RESULTS: The standard MoCA cut-off of 25/26 demonstrated a sensitivity of 88% and a specificity of 51%, resulting in a false positive rate of 20%. ROC analysis showed that a MoCA cut-off of 24/25 has a better balance between sensitivity (81%) and specificity (62%), with a lower false positive rate of 16%. NTB results showed that the MCI group had the lowest norm-referenced percentile scores in the visuo-construction domain, a known early feature of Alzheimer's disease and a significant predictor of a rapid rate of disease progression. CONCLUSIONS: MoCA as a screening tool in individuals ≥65 with type 2 diabetes, overestimates the prevalence of MCI, even when applying lower cut-offs. More specific screening strategies are necessary, particularly targeting the visuo-construction domain, to effectively identify cognitive impairment in individuals with type 2 diabetes.