We aimed to estimate the prevalence of mild cognitive impairment (MCI) among elderly non-responders to a community-based survey. We conducted a two-phase, population-based cross-sectional study of community-dwelling individuals aged 65 years or older in Tone, located in central Japan. The first phase of the study consisted of physical and cognitive examinations of individuals who responded to the first recruitment (quick-responders), whereas the second phase included individuals who did not respond in the first phase (delayed-responders). We compared the prevalence of MCI and dementia between delayed-responders and quick-responders. Of the 2,698 potential candidates, 1,888 (1,619 quick-responders, 225 delayed-responders, and 44 nursing home residents) were enrolled (70.0%). The prevalence of MCI was 2.3-fold increased in delayed-responders compared to the quick-responders (OR=2.27, 95% CI: 1.37-3.77, p=0.002, aged< or =74). In order to develop a method for the early detection of dementia, we must pay more attention to delayed-or non-responders.