Background: The objective of this study was to establish a prospective population-based cohort to investigate the prevalence, incidence and risk factors for mild cognitive impairment (MCI) and dementia. Methods: The Olmsted County, Minn., population, aged 70–89 years on October 1, 2004, was enumerated using the Rochester Epidemiology Project. Eligible subjects were randomly selected and invited to participate. Participants underwent a comprehensive in-person evaluation including the Clinical Dementia Rating Scale, a neurological evaluation and neuropsychological testing. A consensus diagnosis of normal cognition, MCI or dementia was made by a panel using previously published criteria. A subsample of subjects was studied via telephone interview. Results: Four hundred and two subjects with dementia were identified from a detailed review of their medical records but were not contacted. At baseline, we successfully evaluated 703 women aged 70–79 years, 769 women aged 80–89 years, 730 men aged 70–79 years and 517 men aged 80–89 years (total n = 2,719). Among the participants, 2,050 subjects were evaluated in person and 669 via telephone. Conclusions: Strengths of the study are that the subjects were randomly selected from a defined population, the majority of the subjects were examined in person, and MCI was defined using published criteria. Here, we report the design and sampling, participation, baseline measures and sample characteristics.