Nuclear family structures are often thought to be essential for the well-being of children. Divorce, the loss of either biological parent, the presence of step-parents, and the practice of polygynous marriage have all been claimed to negatively impact child well-being. However, empirical research on these topics has been limited by the routine use of cross-regional and cross-sectional databases. Cross-regional data render research vulnerable to the ecological inference fallacy, and cross-sectional data prevent assessment of age-specific impacts of time-varying family-structure variables. When longitudinal data are available, they tend to be drawn from Western/urban contexts. Detailed data on family structure and children’s well-being are rarely collected in more marginalized communities. In many rural and traditional communities, nonnuclear family structures are indeed prevalent and viewed as socially permissible—and, as such, may have different impacts on children’s well-being than in Western contexts. Here, we draw on a detailed, longitudinal dataset from a 20-y prospective study in rural Tanzania, where polygyny and serial monogamy are common. We analyze survival outcomes for 3,693 children born between 1931 and 2014, growth outcomes for 881 children born between 1976 and 2014, and educational outcomes for 1,370 children born between 1976 and 2014. Our analyses indicate that monogamous marriage is not consistently associated with better outcomes for children—contrary to some popular and public health perspectives on human family structure.