Objectives: It is estimated that the proportion of the elderly population ＂who live only as a couple＂ and ＂who live alone＂ will continue to rise. Studies which focus on the dynamic changes in living arrangements and their temporal association with the use of medical services are few. The aim of this study was to fill this gap in knowledge. Methods: The data were drawn from the ＂Survey of Health and Living Status of the Elderly in Taiwan＂, a population-based, longitudinal study of a nationally representative random sample of adults aged 60 years of age and older. The latent growth curve model was applied to explore the longitudinal development and interrelationships between dynamic changes in living arrangements and medical utilization. Results: The cumulative frequency of changes in living arrangements only increased the risk of additional days of hospitalization. A change in living arrangements to living alone reduced the use of outpatient and emergency services, but increased the relative risk of hospitalization and number of days of hospitalization. Remaining as a couple significantly reduced the risk of long-term hospitalization. Conclusions: This study was distinctive in two ways. First, it compared simple and complex households. Second, it focused on whether people were residing as a couple or not. Future public health strategies must take the impact of dynamic changes in living arrangements on medical service into account.