The literature has traditionally found no or little aggregate impact of foreign aid on infant mortality. We suggest that exceptionally high persistence in infant mortality data and sector aid data incompleteness may have played some role in generating such disappointing results. Accounting for both issues, this paper estimates the aggregate impacts of total and sector aid on the neo-natal, infant and under-five mortality rates using fractional estimation techniques for panel data controlling for time-invariant country-specific effects, measurement errors and endogeneity. We confirm that total aid has no impact on child mortality rates. We find mixed evidence that health aid reduces child mortality but robust evidence that agricultural aid has large effects. Aid policies aimed at reducing child mortality in developing countries should recognise the increased importance of targeting the agriculture.