Based on cross-country panel datasets, we find that (i) an increase in population share in agriculture is associated with poverty reduction once the longer-term poverty change or the dynamic is taken into account; (ii) rural non-agricultural sector also is poverty reducing in some cases; and (iii) increased population in the mega cities has no role in poverty reduction. In fact, the growth of population in mega cities is “poverty-increasing” in a few cases. Given that a rapid population growth or rural–urban migration is likely to increase poverty, more emphasis should be placed on policies that enhance support for rural agricultural and non-agricultural sectors. If our analysis has any validity, doubts are raised about recent research emphasising the role of secondary towns or urbanisation as the main driver of extreme poverty reduction.