Social assistance programs often provide a per-person benefit to targeted beneficiaries. Such uniform transfer values do not account for variation in structural deprivations faced by households. In this study, we assess the degree to which a food voucher program for refugees in Lebanon adequately meets the food needs of female- and male-headed households. Using a natural experiment in which some households received an unconditional cash transfer in addition to the food voucher, we analyze spending on food, food consumption, and food coping behaviors that results from the additional cash transfer. We use a regression discontinuity design that estimates program effects among a sample of households that have been assessed as equally needy by implementing agencies. The food voucher program increases food purchases, consumption, and dietary diversity, and reduces food coping strategies. Households who receive the additional cash transfer continue spending more on food and continue to increase food consumption. These latter effects are concentrated in female-headed households, indicating that the food voucher benefit level fell short in providing for these families’ nutritional needs despite the fact that they were assessed as equally impoverished by a proxy means test used to target the program. These results imply that social assistance programs concerned with addressing a specific type of deprivation could take into account structural differences in the incidence of that deprivation when setting benefit levels.