BackgroundLittle is known concerning the lifestyle habits and health conditions in community-dwelling elderly who do not get medical care. We investigated the cross-sectional association between medical expenses (ME) and intellectual activity (IA) in community-dwelling older Japanese.MethodsSelf-administered questionnaires were mailed to all residents born between 1945 and 1949 and covered by A City’s medical insurance system (n = 19,354). Independent variables including health behaviors, oral health, social capital, neighborhood environment, and physical and mental functioning were included in the questionnaires. Medical fee receipts were used to evaluate ME for fiscal 2014, and respondents were classified into no, low, medium, and high ME groups. Higher-level functional capacity was evaluated using the Tokyo Metropolitan Institute of Gerontology Index of Competence, which is comprised of three subscales: instrumental activities of daily living, IA, and social role. Poisson regression models were used to examine the association of ME with IA, with the low ME group as reference.ResultsQuestionnaires were returned by 12,747 individuals (response rate 65.9%). The no ME group had the lowest response rate, the worst lifestyle behaviors, and the lowest social capital, but no problems with neighborhood environment. Higher-level functional capacity, especially IA, was reduced in both the high ME and no ME groups. After adjustments for age, gender, health insurance, accessibility to public facilities in their residential area, family size, body mass index, and physical and mental functioning, the prevalence ratio (PR) for impaired IA lost its significance in the high ME group (PR 0.97, 95% confidence interval 0.90–1.05), but remained significant in the no ME group (1.19, 1.08–1.31). After additional adjustments for health behaviors (i.e., health checks, smoking, fitness, and dietary variety), the PR of the no ME group was attenuated towards the null (1.08, 0.98–1.20).ConclusionsCommunity-dwelling elderly who did not seek medical treatment were indifferent to health surveys and health-promoting behaviors, and undesirable health behaviors were a possible determinant of their impaired IA. Further longitudinal research is needed to confirm the causal associations.Electronic supplementary materialThe online version of this article (10.1186/s12199-017-0672-1) contains supplementary material, which is available to authorized users.