Late-life development may imply terminal processes related with time-to-death rather than with chronological age. In this study, we applied the time-to-death perspective to affective well-being. Using a 15-year observational interval including five measurement occasions with a large sample of deceased participants (N = 1,671; mean age = 75.60; mean time-to-death = 6.83 years, at first occasion) from the Longitudinal Aging Study Amsterdam (LASA), we examined (1) whether intraindividual trajectories in positive (PA) and negative affect (NA) can be better described in terms of time-to-death as compared to chronological age; (2) whether time-to-death-related change in PA and NA follows a terminal decline (linear change) versus terminal drop (accelerating change) pattern, and (3) whether transition points in PA and NA, marking the beginning of the terminal change period, can be identified. For both PA and NA, multilevel mixed models of linear and quadratic time-to-death and age-related trajectories confirmed that time-to-death accounts for more intraindividual variance and reveals a better fit to the data than chronological age. Quadratic time-to-death-related trends fitted better than linear trajectories, thus indicating terminal drop of affective well-being, in terms of decreases of PA and increases of NA accelerating as death comes close. Also, change of NA was linked to time-to-death more strongly than PA change. Transition points to the terminal phase were found at 5.6 and 3.7 years for PA and NA, respectively. In conclusion, affective development toward the end of the human life span may be better understood as a death-related-rather than age-graded-process.