Acemoglu and Johnson (2007) present evidence that improvements in population health do not promote economic growth. We show that their result depends critically on the assumption that initial health has no causal effect on subsequent economic growth. We argue that such an effect is likely, primarily because childhood health affects adult productivity. In our augmented model, which includes initial health, the instrumental variable proposed by Acemoglu and Johnson has no significant predictive power for improvements in health and does not identify the effect of contemporaneous improvements in health on economic growth.