The first stage of sampling was the selection of countries. This selection was stratified according to the WHO regions and the levels of under-five child and adult mortality. Fourteen sub-regions constituted the sampling frame for the first
stage of selection. From each sub-region, a total of four countries were selected at random for participation in the study, with probability proportional to the country population. When the total number of countries was less than four in any sub-region, all the countries were included. This process resulted in 12 sub-regions having four countries each and two sub-regions having three countries each. A total of 54 countries were initially pre-selected, but, due to operational and budgetary reasons, the WHO Global Survey project was implemented in 24 countries from Africa, Asia and Latin America. The second stage of sampling consisted of random selection of two provinces/states (with probability also
proportional to the population size), in addition to the capital city in these 24 countries. A third sampling stage
(also based on the population size but reaching sub-divisions below the province/state level) was used for very large provinces/states. For very large cities (e.g. Mexico City and Beijing)