Global population maps can be valuable for quantifying populations at risk, such as those near nuclear power plants (Nature 472, 400–401; 2011). But the uncertainties inherent in such data sets must be acknowledged. The census data used in map construction for rich countries are recent and detailed. The same is often not true for poorer countries.

For example, Angola's last census was in 1970, broken down into just 18 districts. Estimates of its current total resident population vary from 13.3 million to 19 million, according to the US Census Bureau and the United Nations, respectively. When such outdated and coarse-resolution data are subject to different modelling assumptions by different groups, it can lead to substantially divergent estimates of population distributions and, consequently, populations at risk.

Uncertainties in and between global population maps should be more widely discussed, and a greater effort made to quantify them. Furthermore, spatially referenced demographic data used in map construction are often scattered across national statistical offices and websites. A centralized, open-access, up-to-date database would benefit many fields that rely on population maps, and would require minimal investment.