Little is known about dietary diversity of children residing in areas of high HIV prevalence. This study examined dietary diversity in 381 children ages 6-24 mo in rural South Africa. Twenty-eight (7.3%) children and 170 mothers (44.6%) were HIV infected. Home visits were conducted weekly and a detailed history of dietary intake obtained. A dietary diversity score was computed based on the weekly consumption of 8 food classes. Low dietary diversity was defined as falling within the lowest quartile of the diversity scale. There were 22,772 child weeks of observation: 1369 for HIV-infected children, 8876 for HIV-uninfected children born to HIV-infected mothers, and 12,527 for HIV-uninfected children born to HIV-uninfected mothers. Low dietary diversity was more common in HIV-infected children [crude odds ratio (OR), 2.59; 95% CI, 1.52 to 4.41) compared with children born to HIV-uninfected mothers. In a multiple logistic regression analysis adjusting for socioeconomic and health status, HIV-infected children had lower dietary diversity (conditional OR, 1.76; 95% CI, 1.06 to 2.94) than HIV-uninfected children. HIV-infected children consumed less in 6 of 8 food classes compared with HIV-uninfected children, with the 2 exceptions being breast milk and formula milk. In rural South Africa, HIV-infected children's diets are significantly less diverse than those of HIV-uninfected children. This may be a factor contributing to increased morbidity and poorer survival in these children.