General well-being is known to deteriorate sharply at the end of life. However, it is an open question howrates of terminal change differ across affective and evaluative facets of well-being and if individualdifference correlates operate in facet-specific ways. We examined how discrete affective states (happy,angry, fearful, sad) and satisfaction with key life domains (health, leisure, family) change as peopleapproach death and how differences in end-of-life trajectories are related to sociodemographic (age,gender, education), physical health (disability, body mass index, physician visits), and psychosocialcharacteristics (perceived control, social orientation, living with a partner). We applied growth models to9-year annual longitudinal data of 864 participants (age at death:M75 years, 41% women) from thenationwide German Socio-Economic Panel (SOEP). Findings revealed commonalities and specificities interminal change: Six of seven facets became increasingly fragile late in life (6 to 35 times steeper terminalchange than age change), but at vastly different rates of change (e.g., steep declines in happiness andsatisfaction with health vs. stability in anger) and at different levels at which changes occurred.Commonalities and differences also emerged for the correlates: Those who perceived more control overtheir lives experienced generally more favorable late-life affect and satisfaction trajectories, whereasother correlates operated in more facet-specific ways. For example, participants living with a partner werehappier and more satisfied with family life throughout their last years, but also reported more fear andsteeper increases in sadness, a picture of bittersweet emotions at the end of life.