In many developed countries, population aging raises a number of issues related to the organization and financing of long-term care. While the determinants of the overall burden and cost of care are well understood, the organization of institutionalized long-term care must meet the needs of the elderly. One way to optimize management is to use information on health problems to assess the infrastructure needed, the qualifications of staff, and the allocation of new entrants. In this research, we determine the typical health profiles of institutionalized elderly using novel longitudinal data from nursing homes in the canton of Geneva, Switzerland. Our data contain comprehensive information on health factors such as impairments of psychological and sensory functions, levels of limitations, and pathologies for 21549 individuals covering the period from 1996 to 2018. First, we perform a spectral clustering algorithm and determine the profiles of the institutionalized individuals. Then, we use multinomial logistic regression to study the effects of the factors that determine these health profiles. Our main findings include eight typical health profiles: the largest group consists of the most “healthy” individuals, who, on average, require the least amount of help with their daily needs and who stay in the institution the longest. We show that, in contrast to age at admission and gender, the limitations and the set of pathologies are relevant factors in determining the profile. Our study sheds light on the typical structures of elderly' health profiles, which can be used by institutions to organize their resources and by insurance companies to derive profile-based products that provide additional insurance coverage in case of special needs.