BACKGROUND: In previous studies, cognitive function in elderly type 2 diabetic patients was evaluated by psychometric tests. These studies have confirmed that P300 event-related potential is an objective way of assessing cognitive function.OBJECTIVE: To analyze the objectivity of P300 for assessment of cognitive function in elderly type 2diabetic patients.DESIGN, TIME AND SETTING: This case-control experiment was performed at the Department of Endocrinology of the Fourth Affiliated Hospital, Guangxi Medical University from January 2004 to December 2006.PARTICIPANTS: Seventy-two patients (38 males and 34 females) with type 2 diabetes mellitus were enrolled in this study. The patients were divided according to those with diabetes alone (diabetes alone group) (n=38) and those with diabetes and cerebral ischemia (diabetes and cerebral ischemia group)(n=34). A further 31 healthy individuals (16 males and 15 females), who received health examinations over the same period, were included as normal controls (normal control group).METHODS: All subjects were assessed by Mini-Mental State Examination (MMSE). Abnormalities in cognitive functions were identified by analyzing the auditory P300 event-related potentials.MAIN OUTCOME MEASURES: Auditory event-related potentials and MMSE scores. Multiple linear regression analysis was conducted using the enter method with the 72 elderly patients with type 2diabetes mellitus. P3 latency, P3 amplitude and N2 latency served as dependent variables. Age, sex,education, course of the disease, glycosylated hemoglobin, and ischemic brain damage were used as independent variables.RESULTS: No significant difference in scores of MMSE was detected between the diabetes alone and normal control groups (P > 0.05). MMSE score was significantly lower in the diabetes and cerebral ischemia group (P < 0.01) than in the normal control group. N2 and P3 latencies of auditory event-related potential were significantly longer, and P3 amplitude was significantly lower in the diabetes alone and diabetes and cerebral ischemia groups (P < 0.01) than in the normal control group. N2 and P3 latencies were significantly longer in the diabetes and cerebral ischemia group than in the diabetes alone group (P <0.01), but amplitude was not significantly different. N2 and P3 latencies were negatively correlated with MMSE score in elderly type 2 diabetic patients (r=-0.421, -0.604; both P < 0.01). P3 amplitude was positively related to the score of MMSE (r =0.517; P < 0.01). P3 latency was positively associated with age, course of disease, glycosylated hemoglobin and ischemic brain damage in elderly type 2 diabetic patients (t=2.186 to 3.490; all P < 0.05). P3 amplitude was negatively correlated with age, course of disease and glycosylated hemoglobin (t=-2.220, -2.491, and -2.024, respectively; all P < 0.05). N2latency was positively correlated with age, course of disease and ischemic brain damage (t=2.946, 2.511,and 2.331, respectively; P < 0.05).CONCLUSION: The course of disease, glycosylated hemoglobin and ischemic brain damage are key influential factors for cognitive impairment in elderly type 2 diabetic patients. The P300 event-related potential is a sensitive index for objective assessment of cognitive impairment in elderly diabetic patients.