Abstract. In answers to unique questions from the National Survey of Families and Households, people reveal their valuations of their options outside of marriage as well as their beliefs about their spouses'options. We use this data to demonstrate several features of household bargaining. First, we document marriages in which one spouse would be happier outside the marriage and the other spouse would be unhappier. This provides a new type of evidence that bargaining takes place. Second, we show that spouses have private information about their outside options, and we estimate a bargaining model that quanti…es the extent of resulting ine¢ ciencies. Third, we incorporate caring preferences and imperfect substitutability of utility into the estimation. Without these features, estimation predicts unrealistically high divorce rates, arising because spouses drive too hard a bargain in the presence of asymmetric information and linear utility. After allowing for interdependent and diminishing marginal utility from marital surplus, both of which are identi…ed by incorporating divorce data, our divorce predictions are reasonable. These results show that agents forego their own utility in order to raise the utility of their spouses, and, in doing so, o¤set much of the ine¢ ciency generated by their imperfect knowledge. In contrast, a social planner with only public information about spouses'outside options reduces welfare considerably by keeping far too many couples together. In sum, we …nd evidence about two key features of marriage -asymmetric information and interdependent utility -which are di¢ cult to identify in most studies of interpersonal relationships.We would like to thank