Objective: Depression has been associated with increased mortality rates, and modifying mechanisms have not yet been elucidated. We examined whether specific subtypes or characteristics of late-life depression predict mortality. Methods: A cohort study including 378 depressed older patients according to DSM-IV criteria and 132 never depressed comparisons. The predictive value of depression subtypes and characteristics on the six-year mortality rate, as well as their interaction with somatic disease burden and antidepressant drug use, were studied by Cox proportional hazard analysis adjusted for demographic and lifestyle characteristics. Results: Depressed persons had a higher mortality risk than non-depressed comparisons (HR = 2.95 [95% CI: 1.41-6.16], p = .004), which lost significance after adjustment for age, sex, education, smoking, alcohol, physical activity, number of prescribed medications and somatic comorbidity. Regarding depression subtypes and characteristics, only minor depression was associated with a higher mortality risk when adjusted for confounders (HR = 6.59 [95% CI: 1.79-24.2], p = .005). Conclusions: Increased mortality rates of depressed older persons seem best explained by unhealthy lifestyle characteristics and multiple drug prescriptions. The high mortality rate in minor depression, independent of these factors, might point to another, yet unknown, pathway towards mortality for this depression subtype. An explanation might be that minor depression in later life reflects depressive symptoms due to underlying aging-related processes, such as inflammation-based sickness behavior, frailty, and mild cognitive impairment, which have all been associated with increased mortality.