Objectives The Mini‐Mental State Examination (MMSE) is widely used as a cognitive screening test for older adults; however, cognitive performance can be affected by age, education, and sample selection bias, including time and place. The aims of this study were to examine the distribution of scores on the Japanese version of the MMSE in an urban community sample and to provide normative data for older Japanese adults. Methods A questionnaire survey was conducted on all residents aged 65 years and older living in an urban district in Tokyo (N = 7682). From among this population, 3000 residents were randomly selected to receive visits from trained nurses and to have their health status checked and their cognitive function examined using the MMSE. Results Of the 2786 eligible residents, the MMSE was administered to 1341 (47%) and successfully completed by 1319 (mean age, 74.4 ± 6.4 years; mean years of formal education, 12.6 ± 2.9). The median score was 28. A total of 143 residents (10.8%) had scores below the traditional 23/24 cutoff point. Younger age and higher education were associated with better performance. Greater variation was seen among the oldest and least educated residents, especially among women. Conclusion The results of this study confirm that age and education affect MMSE scores. To ensure the effective use of the MMSE, it is recommended to examine scores corresponding to age and education. The normative data presented are expected to be useful for assessing MMSE scores in older individuals both in and out of the clinical setting. Copyright © 2016 John Wiley & Sons, Ltd.