Recent research has established that the well-being of older adults is very much influenced by the quality and the quantity of their social relationships and their social support network. Weiss has argued that for adults in general, different relationships are needed to meet the diverse requirements of well-being. In this study the Weiss model was evaluated by examining the influence on well-being of the presence versus absence of a confidant and also the presence versus absence of a peer group of social friends. Home interviews were conducted with a representative sample of 100 widowed, divorced, or separated elderly subjects. The interview measured the presence and significance of any confidant relationship, the existence of a peer group, and well-being measured in terms of morale and fulfillment of relational needs. Persons with a confidant and peer group showed a higher well-being score than those with only a confidant. There were no prominent differences between those with peer group only and confidant only relationships. An intervention program based on the Weiss model was developed in which elderly clients were visited on a weekly basis by a volunteer. Volunteers were trained as either relation builders in the Weiss model or were trained as visitors without the relational strategy training. Statistical analysis did not demonstrate superiority of the relational strategy training, although anecdotal evidence supported it. The study appears to have given limited support to the Weiss model. (ABL)