Recent evidence linking premarital cohabitation to high rates of divorce poses a complex theoretical and empirical puzzle. We develop hypotheses predicting that premarital cohabitation is selective of those who are prone to divorce as well as hypotheses predicting that the experience of premarital cohabitation produces attitudes and values which increase the probability of divorce. Using multiwave panel data from a recent cohort of young men and women in the United States, we specify and test models of these predictions. The results are consistent with hypotheses suggesting that cohabitation is selective of men and women who are less committed to marriage and more approving of divorce. The results also are consistent with the conclusion that cohabiting experiences significantly increase young people's acceptance of divorce.