Using data from telephone interviews conducted during 1985, we investigated the prevalence and correlates of depressive symptoms in a national probability sample of 1,232 noninstitutionalized U.S. residents 65 years of age and older. The association between depressive symptoms and personal attributes, personal resources, illness behavior, life events, and self-reported health was examined through chi-square and logistic regression analyses. We measured depressive symptoms with the Center for Epidemiologic Studies--Depression (CES-D) scale using a score of 16 or greater as an indication of high depressive symptoms. Prevalence of high depressive symptoms was 9.9% in the total sample. For black males the prevalence of depressive symptoms was 7.4%, and for white males it was 6.8%. For black females the prevalence of depressive symptoms was 20.8%, while for white females it was 11.5%. In the regression analysis, female gender, single marital status, poor self-reported assessment of health, illness behavior, and a low number of club/organization memberships were significantly associated with high depressive symptoms. In comparison to respondents reporting good to excellent health, those reporting poor to fair health were almost four times more likely to report a high number of depressive symptoms (odds ratio = 3.97).