For a given prevalence, HIV has a relatively higher impact on child mortality when mortality from other causes is low. To project the effect of the epidemic on child mortality, it is necessary to estimate a realistic schedule of "net" age-specific mortality rates that would operate if HIV were the only cause of child death observable. We assume that this net pattern would be independent of mortality from other causes. We used African studies that measured the survival of HIV-infected children (direct data) or survival of children of HIV-infected mothers (indirect data). We developed a mathematic procedure to estimate the mortality of infected children from indirect data sources and obtained net HIV mortality patterns for each study population. The net age-specific HIV mortality pattern for infected children can be described by a double Weibull curve fitted to empiric data; this gives a functional representation of age-specific mortality rates that decline after infancy and rise in the preteens. The fitted curve that we would expect if HIV were the only effective cause of death shows 67% net survival at 1 year and 39% at 5 years. The curve also predicts 13% net survival at 10 years using constraints based on survival of infected adults.