The relationship between women's employment and the risk of divorce is both complex and controversial. The role specialization (or interdependence) view of marriage argues that the gains to marriage for both partners decrease when both are in the labor force, and hence women's employment destabilizes marriage. In contrast, the economic opportunity hypothesis asserts that female labor force participation does not intrinsically weaken marriage, but gives women resources that they can use to leave unsatisfactory marriages. Here we use data from the two waves of the National Survey of Families and Households to conduct the first large-scale empirical test of those conflicting claims. Our results provide clear evidence that, at the individual level, women's employment does not destabilize happy marriages but increases the risk of disruption in unhappy marriages.