A good understanding of the health heterogeneity of elderly people, their characteristics, patterns of health care utilization and subsequent expenditures is necessary to adequately evaluate the policy options and interventions aimed at improving quality and efficiency of care for older people. This article reviews studies that used Latent Class Analysis to identify four health profiles among elderly people in Taiwan: High Comorbidity (HC), Functional Impairment (FI), Frail (FR), and Relatively Healthy (RH). Variables associated with increased likelihood of being in the FR group were older age, female gender, and living with one's family, and these also correlated with ethnicity and level of education. The HC group tended to use more ambulatory care services compared with those in the RH group. The HC group tended to be younger, better educated, and was more likely to live in urban areas than were people in the FI group. The FI group, apart from age and gender, was less likely be of Hakka ethnicity and more likely to live with others than were individuals in the RH group. The FI group had relatively high probabilities of needing assistance, and the FR group had higher healthcare expenditures. A person-centered approach would better satisfy current healthcare needs of elderly people and help forecast future expenditures.