National poverty lines vary greatly across the world, from under $1 per person per day to over $40 (at 2005 purchasing power parity). What accounts for these huge differences, and can they be understood within a common global definition of poverty? For all except the poorest countries, the absolute, nutrition-based, poverty lines found in practice tend to behave more like relative lines, in that they are higher for richer countries. Prevailing methods of setting absolute lines allow ample scope for such relativity, even when nutritional norms are common across countries. Both macro data on poverty lines across the world and micro data on subjective perceptions of poverty are consistent with a weak form of relativity that combines absolute consumption needs with social-inclusion needs that are positive for the poorest but rise with a countrys mean consumption. The strong form of relativism favored by some developed countries -- whereby the line is set at a fixed proportion of the mean -- emerges as the limiting case for very rich countries.