OBJECTIVE - The effect of diabetes on mild cognitive impairment (MCI) and its conversion to dementia remains controversial. We sought to examine whether diabetes and pre-diabetes are associated with MCI and accelerate the progression from MCI to dementia. RESEARCH DESIGN AND METHODS - In the Kungsholmen Project, 963 cognitively intact participants and 302 subjects with MCI (120 with amnestic MCI [aMCI ] and 182 with other cognitive impairment no dementia [oCIND]) age ≥75 years were identified at baseline. The two cohorts were followed for 9 years to detect the incident MCI and dementia following international criteria. Diabetes was ascertained based on a medical examination, hypoglycemic medication use, and random blood glucose level ≥11.0 mmol/l. Pre-diabetes was defined as random blood glucose level of 7.8 -11.0 mmol/l in diabetes-free participants. Data were analyzed using standard and time-dependent Cox proportional-hazards models. RESULTS - During the follow-up period, in the cognitively intact cohort, 182 people developed MCI (42 aMCI and 140 oCIND), and 212 developed dementia. In the MCI cohort, 155 subjects progressed to dementia, the multi-adjusted hazard ratio (95% CI) of dementia was 2.87 (1.30-6.34) for diabetes, and 4.96 (2.27-10.84) for pre-diabetes. In a Kaplan-Meier survival analysis, diabetes and pre-diabetes accelerated the progression from MCI to dementia by 3.18 years. Diabetes and pre-diabetes were neither cross-sectionally nor longitudinally associated with MCI. CONCLUSIONS - Diabetes and pre-diabetes substantially accelerate the progression from MCI to dementia, and anticipate dementia occurrence by more than 3 years in people with MCI. The association of diabetes with the development of MCI is less evident in old people. © 2010 by the American Diabetes Association.