Aim To investigate which category in the “Kihon Checklist” developed by the Japanese Ministry of Health, Labor and Welfare can predict functional decline for community‐dwelling elderly people at a 2‐year follow up. Method We compared comprehensive geriatric assessment (CGA) between “specified elderly individuals” at high risk of requiring long‐term care insurance (LTCI) and “uncertified elderly people” (neither certified under LTCI nor “specified”), and also compared CGA between the risk group and non‐risk group, in subcategories of the “Kihon Checklist”, such as physical strength, nutrition/oral function, overall low score on questions 1–20, houseboundness, cognitive function, and depression risk. The study population consisted of 527 elderly participants aged 75 years and older in a cross‐sectional study, and 382 in a longitudinal study. CGA was assessed for basic and higher functional activities of daily living (ADL), depressive symptoms, and quality of life (QOL). The Student's t‐test was used in the cross‐sectional study and ANOVA with repeated measures was used in the longitudinal analysis. Results In the cross‐sectional study, the risk group had lower functions in all CGA items than the non‐risk group in all subcategories of the “Kihon Checklist.” In the longitudinal study, Tokyo Metropolitan Institute of Gerontology Index of Competence scores and its three subscales declined in the risk group both in physical and cognitive subcategories compared with the non‐risk group, whereas only one or two subscales of Tokyo Metropolitan Institute of Gerontology Index of Competence declined in “specified” and the other two subcategories of the Kihon Checklist Conclusion In both cross‐sectional and longitudinal studies, the assessment of physical strength and cognitive function was more useful to detect frail elderly. Geriatr Gerontol Int 2013; 13: 654–662.