Studies of the effect of depression on life expectancy have found controversial results. In the past, an excess mortality was attributed to conditions seen at psychiatric institutions. More recently, in addition to an excess mortality due to violent causes of death, some studies also found excess mortality due to non-violent and all-cause mortality. Hospital studies that evaluated this relationship are generally flawed and lack of control for potential confounders. Underlying physical illnesses could explain the excess mortality reported. On the other hand, community-based studies have studied small number of depressed subjects, and have used diverse measurements of depression which hamper comparability. This paper critically reviews recent published studies and discusses possible mechanisms for this association.