ABSTRACT. Exchange theory assumes that people prefer balanced support exchanges in their relationships. If there is an imbalance and no expectation of change in the future, a relationship might be terminated. The question is which relationships are discontinued. The data are from a longitudinal study of 2,057 older adults who identify 18,915 relationships at T1. A relationship is regarded as discontinued if it is not identified by the older adult at the second and third measurement moments. Of the T1 relationships, 4,042 have since been discontinued. The results of a multilevel logistic regression analysis show that the more intensive the support exchanges are at T1, the more likely it is for relationships to be continued. Relationships where older adults are overbenefited with instrumental support, i.e. receive more than they give, have a higher chance of being continued. However, if older adults are overbenefited with emotional support, this decreases the chance of the relationships continuing. The type of relationship has a significant effect on whether or not it is continued. Close kin relationships are most likely to be continued, and relationships with less close kin, friends and neighbors have a higher chance of being discontinued. The costs of the relationship are also decisive; the higher the contact frequency and the lower the traveling time to the network member, the higher the chance of the relationship being continued. Furthermore, the larger the network of the older adult, the more likely it is for an unbalanced relationship to be discontinued.
