This paper uses data from Matlab, Bangladesh to examine the characteristics of female-headed households and estimate the impact of female-headship on children's schooling. Female householdheads in Matlab fall into two broad groups: widows and married women, most of whom are wives of migrants. These women differ from each other not only in their current socio-economic circumstances, but also in their backgrounds and circumstances prior to getting married. To identify the effects of female-headship on children's outcomes, I use a two-stage least squares strategy that controls for the possible endogeneity of both types of female-headship. Results indicate that children residing in households headed by married women have stronger schooling attainments than children in other households, while children of widows are more likely to work outside the home. The hypothesis of exogeneity of female-headship is rejected in most cases.