Mild cognitive impairment (MCI) is proposed to be a prodrome to dementia in some older adults. However, the presentation of MCI in the community can differ substantially from clinic-based samples. The aim of the current study is to demonstrate the effects of different operational definitions of MCI on prevalence estimates in community-dwelling older adults. A consecutive series of 200 participants aged 65 and over from the Adult Changes in Thought (ACT) community-based cohort were approached to undergo comprehensive neuropsychological and medical evaluation; 159 were included in the final analyses. Nondemented subjects were categorized using various diagnostic criteria for MCI. In a novel approach, neuropsychological test scores were evaluated using an individualized benchmark as a point of test comparison, as well as traditional methods that entail comparison to age-based normative data. Diagnostic criteria were further subdivided by severity of impairment (1.0 vs. 1.5 standard deviations [sd] below the benchmark) and extent of impairment (based on a single test or an average of tests within a cognitive domain). MCI prevalence rates in the sample were highly dependent on these diagnostic factors, and varied from 11% to 92% of the sample. Older groups tended to show higher prevalence rates, although this was not the case across all diagnostic schemes. The use of an individualized benchmark, less severe impairment cutoff, and impairment on only a single test all produced higher rates of MCI. Longitudinal follow-up will determine whether varying diagnostic criteria improves sensitivity and specificity of the MCI diagnosis as a predictor for dementia.