This study combines stress theory and the life course perspective to examine patterns and predictors of depressive symptoms in old age. Much of the research on depressive symptoms examines patterns across the life course, treating the elderly as a single category. While this perspective demonstrates important contrasts across the life course, it tends to ignore important variation that occurs within the age range typically considered to be the elderly (Feinson, 1985; Newmann, 1989; Newmann, Engel, and Jensen, 1991). Stress theory and the life course perspective provide a strategic context for studying diverse relationships among self-reported physical health, household income, social support, sense of control, and depressive symptoms for people age 55 or older;Data from telephone interviews with nearly 1,000 elderly respondents are used to examine age variation in patterns and predictors of depressive symptoms among three groups of older people: the pre-retired old (age 55 to 64), the young old (age 65 to 74), and the very old (age 75 or older). Hypotheses related to differential exposure, differential vulnerability, and differential resources are tested. The very old are found to have significantly higher levels of depressive symptoms than the pre-retired old. For each of the elderly age groups, health perception is significantly related to depressive symptoms. While the very old are more likely to report poorer health, the impact of health on depressive symptoms does not vary significantly from that of the other two age groups, after controlling for socioeconomic and resource variables. Thus, greater exposure to health problems, rather than greater vulnerability, is the important predictor of mental health outcomes. Having fewer economic, social, and psychological resources, particularly psychological resources, also plays an important role in depressive symptoms of the very old. Based on these findings, this study increases our understanding of the complicated relationships among stressors, resources and depressive symptoms for three groups of older people. This is accomplished by demonstrating that the life course of the elderly is diverse and heterogeneous. Consequently, it is very difficult to make any simple generalizations about their lives, including their mental health.