This study applies multilevel logistic regression to Demographic and Health Survey data from 22 sub‐Saharan African countries to examine whether the relationship between child mortality and family structure, with a specific emphasis on polygyny, varies cross‐nationally and over time. Hypotheses were developed on the basis of competing theories on the relationship between child health and family structure. Although children of mothers in polygynous marriages are more likely to die than those of mothers in monogamous unions, the relationship is constant across time. Familial factors including education, socioeconomic status (SES), and urban residence accounted for most of the observed cross‐national variation associated with polygyny. Consequently, improving maternal education and household SES would greatly benefit child health in sub‐Saharan Africa.