Aid agencies often provide transfers in food rather than cash out of a paternalistic belief that food transfers will better improve household food security. However, evidence from Latin America shows that cash transfers often increase the share of food in consumption, counter to Engel’s Law. This finding suggests households treat transfer income differently, with previous literature arguing that transfers shift intrahousehold bargaining. Until now, there has been little rigorous evidence on how the effects of transfers on food consumption patterns differ by context or by transfer modality. We use experimental data from three countries, Ecuador, Uganda and Yemen, to test the relative impact of food transfers and cash transfers (and vouchers in Ecuador) on the food share of consumption, food Engel curves, and the composition of food consumption. We find that, in all three countries, there are no significant differences by modality in the impacts of transfers on overall food share or food Engel curves. In particular, in Ecuador, transfers in the form of food, cash, or vouchers all increase the share of food in total consumption, representing outward shifts of the food Engel curves, but there are no significant differences by modality in these shifts. In Uganda, neither food nor cash significantly changes the food share of consumption. In Yemen, there is also no significant difference in Engel curves between food and cash beneficiaries. However, we find in all three countries that there are differences by modality in impacts on the composition of food consumption. In two of the three countries, food transfers lead to increased food group shares of the items included in the food ration, and in all three countries, cash transfers lead to larger improvements in dietary diversity. We find no evidence of changes in intrahousehold bargaining power due to any transfer modality in all three countries, suggesting another factor may be responsible for shifting Engel curves in Ecuador.