The association between the presence of depressive symptoms and cancer incidence and mortality and mortality from noncancer causes was studied in a population-based cohort of 6848 persons free of cancer who were followed from 1965 to 1982 as part of the Alameda County study. Age-adjusted and multivariate analyses involving over 111,000 person-years of follow-up demonstrated an association between high levels of depressive symptoms at baseline and deaths from noncancer causes but no association with either cancer incidence or cancer mortality. Our analyses suggest the possibility that the presence of previously diagnosed cases of cancer and the inclusion of items which tap somatic problems in depression scales may contribute to differences between these results and others in which depression has been linked to cancer mortality.