The role of duration of depressed mood in the prediction of cardiovascular disease (CVD) requires further study, as it has been suggested that emerging depressive symptoms may be a better predictor than persistent depressive symptoms. This prospective cohort study of 3,701 men and women aged > 70 years uses 3 measurement occasions of depressive symptomatology (Center for Epidemiologic Studies-Depression Scale) during a 6-year period to distinguish persons who were newly (depressed at baseline but not at 3 and 6 years before baseline) and chronically depressed (depressed at baseline and at 3 or 6 years before baseline). Their risk of subsequent CVD events and all-cause mortality was compared with that of subjects who were never depressed during the 6-year period. Outcome events were based on death certificates and Medicare hospitalization records. During a median follow-up of 4.0 years, there were 732 deaths (46.2/1,000 person-years) and 933 new CVD events (64.7/1,000 person-years). In men, but not in women, newly depressed mood was associated with an increased risk of CVD mortality (relative risk 1.75, 95% confidence interval [CI] 1.00 to 3.05), new CVD events (relative risk 2.07, 95% CI 1.44 to 2.96), and new coronary heart disease events (relative risk 2.03, 95% CI 1.28 to 3.24) after adjustment for traditional CVD risk factors. The association between newly depressed mood and all-cause mortality was smaller (relative risk 1.40, 95% CI 0.95 to 2.07). Chronic depressed mood was not associated with new CVD events or all-cause mortality. Our findings suggest that newly depressed older men, but not women, were approximately twice as likely to have a CVD event than those who were never depressed. In men, recent onset of depressed mood is a better predictor of CVD than long-term depressed mood.