The objective of this research was to assess the association of child mortality with polygyny and maternal HIV status through a prospective community-based study in Rakai district, Uganda. We sought to test whether there was an indirect evidence that polygynous households in an HIV prevalent area may divert resources away from the children of HIV-infected mothers in favor of children with better survival prospects. We test this theory using data from a follow-up study which collected detailed behavioral and medical information at 10-month intervals on a cohort of over 4000 pregnant women and their infants (5300 person years of observation). Cox proportional hazards models estimated the mortality hazard (RR) associated with polygyny for children of HIV-negative and HIV-positive mothers. HIV prevalence in the full cohort of mothers was 11.9%, and 23% of mothers lived in polygynous households. Multivariate analysis showed an increased hazard of child mortality if the mother was HIV-positive (RR = 1.75, p<0.001). Maternal education reduced mortality, whereas low birth weight increased mortality risk. Polygyny was associated with an increase in the hazard of child mortality in the full sample (RR = 1.36, p<0.001) and in mothers who were HIV-positive (RR = 2.17, p<0.001), but not in HIV-negative mothers. Being born to an HIV-positive mother increased mortality risk and polygyny accentuated a child's risk of death. Polygyny had no significant effect on the survival of children with HIV-negative mothers. Polygynous households, where not all wives may have HIV, could be diverting resources away from the children of the infected wives.