Type 2 diabetic (T2D) subjects have a significantly higher risk of developing mild cognitive impairment (MCI) and dementia than age-matched non-diabetic individuals. However, the accurate evaluation of cognitive status is based on complex neuropsychological tests, which makes their incorporation into the current standard of care for the T2D population infeasible. Given that the ability to maintain visual gaze on a single location (fixation) is hampered in Alzheimer's disease (AD), the aim of the present study was: (1) To assess whether the evaluation of gaze fixation during fundus-driven microperimetry correlated with cognitive status in T2D subjects; (2) to examine whether the addition of fixational parameters to the assessment of retinal sensitivity increased the predictive value of retinal microperimetry in identifying T2D subjects with MCI. For this purpose, fixation parameters and retinal sensitivity were compared in three age-matched groups of T2D subjects: normocognitive (n = 34), MCI (n = 33), and AD (n = 33). Our results showed that fixation is significantly more unstable in MCI subjects than normocognitive subjects, and even more altered in those affected by AD (ANOVA; p < 0.01). Moreover, adding fixation parameters to retinal sensitivity significantly increases the predictive value in identifying those subjects with MCI: ROC (Receiver Operating Characteristic) Area 0.68 with retinal sensitivity alone vs. ROC Area 0.86 when parameters of fixation are added to retinal sensitivity (p < 0.01). In conclusion, our results suggest that fixational eye movement parameters assessed by fundus-microperimetry represent a new tool for identifying T2D subjects at risk of dementia.