Older people engage in social interaction less frequently than their younger counterparts. As I mentioned at the start, the change has been interpreted in largely negative terms. Yet when asked about their social relationships, older people describe them as satisfying, supportive, and fulfilling. Marriages are less negative and more positive. Close relationships with siblings are renewed, and relationships with children are better than ever before. Even though older people interact with others less frequently than younger people do, old age is not a time of misery, rigidity, or melancholy. Rather than present a paradox, I argue here that decreasing rates of contact reflect a reorganization of the goal hierarchies that underlie motivation for social contact and lead to greater selectivity in social partners. This reorganization does not occur haphazardly. Self-definition, information seeking, and emotion regulation are ranked differently depending not only on past experiences, but on place in the life cycle and concomitant expectations about the future. I contend that the emphasis on emotion in old age results from a recognition of the finality of life. In most people's lives this does not appear suddenly in old age but occurs gradually across adulthood. At times, however, life events conspire to bring about endings more quickly. Whether as benign as a geographical relocation or as sinister as a fatal disease, endings heighten the salience of surrounding emotions. When each interaction with a grandchild or good-bye kiss to a spouse may be the last, a sense of poignancy may permeate even the most casual everyday experiences. When the regulation of emotion assumes greatest priority among social motives, social partners are carefully chosen. The most likely choices will be long-term friends and loved ones, because they are most likely to provide positive emotional experiences and affirm the self. Information seeking will motivate some social behavior, but for reasons discussed previously, this will also require judicious choices of social partners. Narrowing the range of social partners allows people to conserve physical and cognitive resources, freeing time and energy for selected social relationships. As such, SST is highly consistent with the selective optimization with compensation model of successful aging formulated by P. Baltes and M. Baltes (1990) described above. SST is meant to describe and explain the underlying mechanisms for age-related changes in social behavior. It is not intended to be prescriptive.(ABSTRACT TRUNCATED AT 400 WORDS)