Working with the Bill and Melinda Gates foundation and UNFPA, WorldPop-Flowminder are producing high resolution maps of the age and sex structure of populations in low income countries. This work has been in support global polio eradication initiative, which aims to deliver polio vaccines to all children under the age of 5. Previous vaccination efforts found that inaccurate estimates of population size and demographic structures resulted in millions of dollars of wasted vaccines as well as hundreds of thousands of children not being vaccinated. Using household survey data collected from across countries, we applied multinomial logit models that take advantage of correlations between gridded environmental data layers and the proportion of individuals in each of 34 age/sex categories. Using this modelling approach, we have been able to predict the numbers of individuals in each age class with up to 78% predictive accuracy. Importantly, we are also able to assess the uncertainty in each age class, allowing us to determine whether predicted differences in different regions are meaningful. These results will be used in conjunction with bottom-up population models to estimate the total number of people in each age class at 100 m resolution across countries including Afghanistan and Nigeria. Aside from supporting the distribution of vaccines. this information will be vital in informing policy on a wide range of applications including understanding health and welfare needs of the population and forecasting economic prospects for different regions of target countries. The data are currently used in the Nigeria national vaccination tracking system: http://vts.eocng.org/.