Do individuals divorce for economic reasons? Can we measure the attractiveness of new matches in the marriage market? We answer these questions using a structural model of the household and a rich panel dataset from Malawi. We propose a model of the household with consumption, production and revealed preference conditions for stability on the marriage market. We define marital instability in terms of the consumption gains to remarrying another individual in the same marriage market, and to being single. We find that a 1 percentage point increase in the wife's estimated consumption gains from remarriage is significantly associated with a 0:6 percentage point increase in divorce probability in the next three years. In a multinomial model, higher values of consumption gains from remarriage raise the odds of divorce and remarriage but not of divorce and singleness. These findings provide out-of-sample validation of the structural model and shed new light on the economic determinants of divorce.