Marital dissolution is commonly assumed to cause increased depression among adults, but causality can be questioned based on directionality and third-variable concerns. The present study improves on past research by using a propensity score matching algorithm to identify a subsample of continuously married participants equivalent in divorce risk to participants who actually experienced separation/divorce between two waves of the nationally representative Midlife Development in the United States (MIDUS) study. After correcting for participants’ propensity to separate/divorce, we observed increased rates of depression at the second MIDUS assessment only among participants who (a) were depressed at the initial MIDUS assessment and (b) experienced a separation/divorce. Participants who were not depressed at the initial assessment but who experienced a separation/divorce were not at increased risk for a later major depressive disorder. Thus, both social selection and social causation contribute to the increased risk for a major depressive disorder found among separated/divorced adults.