Data from both spouses in a population-based sample of middle-aged and older adults (N = 1,869 couples) were used to evaluate the 2-year prospective association between marital discord and depressive symptoms. Nested path analyses indicated that (a) baseline marital discord predicted one's own depressive symptoms at follow-up, (b) baseline depressive symptoms predicted one's own marital discord at follow-up, (c) baseline depressive symptoms predicted partners' marital discord at follow-up, and (d) there were no gender differences in the magnitudes of the pathways between one's own baseline depressive symptoms and one's own marital discord at follow-up or between one's own baseline marital discord and one's own depressive symptoms at follow-up. These results suggest a bidirectional longitudinal association between marital discord and depressive symptoms in middle-aged and older adults.