Death becomes exponentially more probable with each year of aging. Acknowledging this when making health care decisions has important consequences for quality of life, as the benefits of care may not outweigh the costs if little time remains to enjoy those benefits. Death expectations across the lifespan influence how people prepare for and experience death, but little is known about the trajectories of death expectations in the final years of life and how those trajectories relate to health care use. Using biennial subjective survival probability (SSP) ratings gathered over the decade prior to death from participants ages 65 and older in the Health and Retirement Study (HRS), latent profile analyses (LPA) were used to identify profiles of participants using meaningful subgroups. The optimal solution included four distinct profiles of SSP trajectories labeled realists, non-commitals, pessimists, and optimists. A series of t-test, chi-square, one-way ANOVA, and repeated measures ANOVA analyses were then conducted to identify characteristic features of the groups and assess for any differences in health care behaviors and death attitudes. Results demonstrated that two of four profiles had a significant increase in SSP from wave four to wave five: the optimists and non-committals. The pessimists were older at death than non-committals and realists, and were more likely to have had a final illness duration of more than a year than the full sample; pessimists had worse self-perceptions of health than optimists and realists; optimists were less likely to have a living will than the full sample and more likely to identify as African American than their representations in the full sample; and optimists reported fewer depressive symptoms and health conditions than pessimists. I also hypothesized that SSPs during the last decade of life would predict health care use in the same period and that optimists and realists would use more health care health care than pessimists; these hypotheses were not confirmed.