BackgroundThe relationship between the duration of depressive symptoms and mortality remains poorly understood.AimsTo examine whether the duration of depressive symptoms is associated with mortality risk.MethodData (n = 9560) came from the English Longitudinal Study of Ageing (ELSA). We assessed depressive symptom duration as the sum of examinations with an eight-item Center for Epidemiologic Studies Depression Scale score of ⩾3; we ascertained mortality from linking our data to a national register.ResultsRelative to those participants who never reported symptoms, the age- and gender-adjusted hazard ratios for elevated depressive symptoms over 1, 2, 3 and 4 examinations were 1.41 (95% CI 1.15–1.74), 1.80 (95% CI 1.44–2.26), 1.97 (95% CI 1.57–2.47) and 2.48 (95% CI 1.90–3.23), respectively (P for trend <0.001). This graded association can be explained largely by differences in physical activity, cognitive function, functional impairments and physical illness.ConclusionsIn this cohort of older adults, the duration of depressive symptoms was associated with mortality in a dose–response manner.