The death of a significant other is seen as a major life disruption, with profound consequences for the bereaved person. This type of event occurs more and more frequently with advancing age. In this chapter, the two main questions we endeavor to answer are: in very old age, what are the main sources of bereavement, and what are the consequences of such a loss on health and on relationships? The bereaved were split according to the category of person being grieved: a close relative or a friend, and then compared with two other groups of non-bereaved elders: one consisting of those who gad declared no recent significant change in their life, and the other of those who had reported serious recent health trouble. This chapter is an upgraded version of a previous paper (Lalive d'Epinay, Calli, & Spini, 2003) which was based on the observation of Swilsoo's first octogenarian cohort over a period of five years (1994-1999). Here we used all data of the completed research, which provided a ten-year follow-up of the first cohort (1994-2004) and a five-year follow-up of the second (1999-2004). Participants on baseline were 652, with a total of 2496 interviews. In the 2003 paper, we used analyses of covariance ({ANCOVA}); here, we apply multilevel analyses which are especially appropriate for repeated data series. In very old age, the great majority of the loved ones who die are either sibliings or friends. The loss of either a close relative or a friend has an adverse impact on the psychological health of the bereaved, but each person displays a distinctive pattern of consequences. The death of a close relative stimulates a tightening of ties whith the family that seems to buffer part of the impact on well-being. No such relational regulation comes into play upon the loss of a friend, with the result that the bereaved person suffers a sharp increase in depressive symptoms; further, the loss of a friend has a delayed disruptive effect of men's sense of closeness.