The traditional per-capita based poverty measurements that ignore the heterogeneity in household size and composition (and hence the potential economies of scale in consumption) in female-headed households (FHHs) and male-headed households (MHHs), appear to underestimate the incidence of poverty of FHHs. Using two waves of Household Income and Expenditure Survey data (2012/2013 and 2016) collected by the Department of Census and Statistics for Sri Lanka, this study examines whether there is a difference between the level of food consumption economies of scale materialised by MHHs and FHHs, and whether FHHs overrepresent the poor after economies of scale in food consumption is incorporated when estimating poverty measurements in Sri Lanka. The Engel curves and implied equivalence scales were used for the empirical analysis. Our results indicate that de-jure FHHs materialise higher economies of scale in food consumption than de-facto FHHs and MHHs, and that the incidence of poverty is higher among de-jure FHHs even after economies of scale for food consumption are allowed for. Adjusting the poverty measurements for consumption economies of scale is critical to identify the poorest of the poor and enhance their welfare.