The association between the size and structure of social networks and the prevalence of depressive symptoms was examined in a population-based study of 1615 men and women age 65 years and older. Age was significantly associated with marital status, social network index quartile, and the social relationship to the primary source of support. Women and men differed with respect to current marital status, number of close friends and relatives, frequency of face-to-face contact, and participation in voluntary associations and religious institutions. Regardless of their marital status, women were also less likely than men to point to a spouse as their primary source of support. Beck Depression Inventory mean scores and rates of depressive symptoms were inversely associated with social network index and participation in voluntary associations and religious institutions for both men and women. Individuals with no primary source of support or who depended on a relative had significantly higher than expected mean scores and rates of depressive symptoms. Both social network index and social distance to primary source of support were independently associated with depression after controlling for age, sex, and number of chronic conditions. Participation in voluntary associations, social distance from primary source of support, church membership, and number of close friends were also significant independent predictors of depressive symptoms. Results indicate that depressive symptoms are inversely associated with the size of social networks. The structure of these networks, in turn, is influenced by biological factors such as age, physical disability, and mortality of network members, and by culturally-determined rules that define the individuals and institutions available for support. However, these rules appear to differ for men and women.