Objective To explore the correlation between mild cognitive impairment (MCI) and elderly type 2 diabetes mellitus (T2DM) and pre-diabetes (impaired glucose regulation, IGR). Methods The elderly T2DM patients (60-80 years old) admitted to our department from January to December 2018 were recruited, and 30 IGR patients and 30 healthy people (control group) for physical examination in the same period were enrolled. The incidence of MCI, general data and related clinical indices were collected and compared among the 3 groups. According to the existence of MCI, the T2DM patients were divided into MCI subgroup (n=39) and non-MCI subgroup (n=21 cases), and IGR patients were also assigned into MCI (n=14) and non-MCI subgroup (n=16). The clinical data were compared between the MCI and non-MCI subgroup in T2DM group and IGR group. SPSS statistics 17.0 was used for data processing. Multivariate logistic regression analysis was employed to explore the independent influencing factors of MCI in elderly patients with T2DM or IGR. Results Compared with that of the control group [30.0%(9/30)], the incidence rate of MCI was significantly higher in the IGR group [46.7%(14/30), P=0.002] and T2DM group [65.0%(39/60), P=0.004], and it was obviously higher in the latter group than the former group (P=0.006). Compared with control group, IGR group and T2DM group had significantly shorter educational length and lower score of Montreal cognitive assessment (MoCA) scale, and increased levels of low-density lipoprotein cholesterol (LDL-C), glycosylated hemoglobin Alc (HbA1c), fasting blood glucose (FBG) and postprandial 2-hour blood glucose (PBG). What′s more, MoCA score was reduced while the levels of HbA1c, FBG and PBG were increased significantly in the T2DM group than the IGR group (P<0.05). Univariate analysis showed that age, systolic blood pressure (SBP), and levels of LDL-C, HbA1c and FBG were significantly higher, while education length was shorter and MoCA score was lower in the MCI subgroups than those of corresponding non-MCI subgroups in both T2DM group and IGR group (P<0.05). Multiple logistic regression analysis showed that age (OR=1.104,5%CI 1.026-1.132, P=0.023; OR=2.001,5%CI 1.296-2.134, P=0.038), FBG (OR=2.138,5%CI 0.283-6.132, P=0.042; OR=0.986,5%CI 0.010-2.448, P<0.001) and LDL-C (OR=3.382, 95%CI 1.613-8.206, P=0.008; OR=2.682,5%CI 1.983-5.751, P=0.022) were the same independent risk factors, while educational years was the same independent protective factors (OR=1.688,5%CI 0.172-3.394, P=0.002; OR=1.692, 95%CI 0.187-3.412, P=0.003) for MCI in the elderly T2DM and IGR patients. Conclusion The elderly T2DM patients are at high risk of MCI. This trend has appeared since the IGR stage. Therefore, it is necessary to control the levels of FBG and LDL-C in elderly patients with IGR as early as possible.