Aim The present study aimed to identify the health profiles of older adults by using latent class analysis to investigate health heterogeneity and to determine what factors predicted healthy aging among an oldest‐old sample cohort that was followed up for 14 years in Taiwan. Methods Data were drawn from five waves (carried out in 1993, 1996, 1999, 2003 and 2007) of the Taiwan Longitudinal Study on Aging to examine the changes in health heterogeneity in a nationally representative oldest‐old cohort of Taiwanese. Overall, data from a total of 11 145 observations of 3155 older adults were considered. The influential factors predicting health changes were analyzed by using a generalized estimating equation. Results The results showed that four health profiles were identified among the aging population observed in the Taiwan Longitudinal Study on Aging. With increasing age, the combined effects of the physical functioning, cognitive and emotional health, and comorbidities of older adults significantly impact their health changes. Apart from health deteriorating with age and sex disparities, educational and economic status, health behaviors, and social participation at the individual level were found to be the robust factors in predicting healthy aging. Conclusions In considering what factors impact healthy aging, we suggest that a person‐centered approach would be useful and critical for policy makers to understand the compositions of health profiles and the influencing factors in view of a life‐course perspective. Based on the factors identified as influencing healthy aging at the individual level, it is imperative from a policy‐making perspective to maximize opportunities for healthy aging. Geriatr Gerontol Int 2017; 17: 697–707.