Depression is associated with an increased risk of mortality in studies that included adults of all ages, but results of studies restricted to older adults are less consistent. This study evaluated the association between depressive symptoms and mortality among 764 white women aged 65+ years in Baltimore, Maryland, and examined methodologic and conceptual issues regarding this association. The Center for Epidemiologic Studies-Depression Scale (CES-D) was administered in face-to-face interviews in 1984. Mortality data were collected through 1990. The 6-year risk of death was 14.5% among women with CES-D scores of 0–1, 24% to 28% among women with scores of 2–24, and 47% among those with scores over 24. The adjusted hazards ratio (RR) comparing women with the highest (25–58) vs lowest (0–1) scores was 1.77 (95% confidence interval (CI) = 0.91–3.42). Depressive symptoms were only weakly associated with mortality when using the CES-D scale dichotomized at the traditional cutpoint of 16 (RR = 1.10, CI = 0.73–1.66), or when the follow-up period was 2 years. Furthermore, depressive symptoms were associated with mortality only among women in poor health. The association between depressive symptoms and mortality risk appeared to be affected by baseline physical health, length of follow-up, and measurement of depression. (Epidemiology 1999;10:54–59)