Income‐differentiated mortality, by reducing the share of poor persons in the population, leads to the “Mortality Paradox”: the worse the survival conditions of the poor are, the lower is the measured poverty. We show that FGT measures (Foster et al., ) are, in general, not robust to variations in survival conditions. Then, following Kanbur and Mukherjee (), we propose to adjust FGT poverty measures by extending the income profiles of the prematurely dead, and we identify the condition under which so‐adjusted FGT measures are robust to mortality changes. Finally, we show, on the basis of data from 2007 on old‐age poverty in 11 European economies, that the effect of extending income profiles of the prematurely dead on poverty measurement varies with: (1) the fictitious income assigned to the prematurely dead; (2) the degree of poverty aversion; (3) the shape of the (unadjusted) income distribution; and (4) the strength of the income/mortality relationship.