The cumulative advantage hypothesis suggests diverging socioeconomic status (SES) based gaps in health with age. However, previous studies yield inconsistent findings regarding the association between SES and health across the adult life span. Dealing with the issue of mortality selection bias, this study utilizes latent growth-curve modeling to comprehensively examine age trajectories of both physical and mental health by SES using panel data based on a national probability sample of 3617 US adults. We find that education- and income-based gaps in physical impairment and the education-based gap in depression diverge over time for all adult age groups, supporting the hypothesis of cumulative advantage. In contrast, we find that the income-based gap in depression converges in older age, supporting the hypothesis of age-as-leveler. Mortality selection bias is unlikely to be a major part of the explanation for the convergence. These results indicate that age-related patterns in health trajectories may differ by various dimensions of SES and health. Finally, we take into account persistence or change in income over time to examine the relationship between trajectories of income and health across adulthood, highlighting the importance of considering the temporal patterns of income in understanding age trajectories of health.