OBJECTIVES: To apply diagnostic criteria for mild cognitive impairment (MCI) to a geographically representative sample, to estimate the prevalence of MCI, and to estimate 10-year trajectories of incident disability for cognitively intact participants and subgroups with MCI. DESIGN: Prospective cohort; 10 years of follow-up. SETTING: Community-based survey of noninstitutionalized population aged 65 and or older in two rural Iowa counties (Washington and Iowa). PARTICIPANTS: Iowa Established Populations for Epidemiologic Studies of the Elderly (aged > or = 65; N = 3,673; 61.3% female; 99.9% white). MEASUREMENTS: Age, sex, education, Short Portable Mental Status Questionnaire (SPMSQ), 20-item word recall, activities of daily living (ADLs), instrumental activities of daily living (IADLs), chronic medical conditions. RESULTS: MCI was prevalent in 24.7% of participants at baseline. Most participants in the overall cohort remained stable or changed slowly (< or = 1 new limitations) over 10 years (63.1% for SPMSQ, 89.3% for word recall, and 61.7% for ADL disability). For MCI/no prevalent IADL disability (Stage 1 MCI), disability progression was similar to that in the cognitively intact subgroup (median = 0.08 vs 0.05 disabilities per year). For MCI plus prevalent IADL disability (Stage 2 MCI), the median rate of change was equivalent to that of the severely impaired (0.23 disabilities per year; interquartile range = 0.12-0.36). CONCLUSION: Unlike participants with MCI who reported no IADL limitations, those with such limitations were more likely to develop ADL disability--a prerequisite for a diagnosis of dementia.