A range of scenarios is presented indicating plausible upper limits on the level of food aid likely to be required by the end of the 1980s. Projections are found to imply large increases in imports of food staples, failing a significant improvement in the longer term trends in domestic production of food in developing countries. However, actual levels of food aid should be lower because of economic conditions and management constraints on the amounts that can be used with positive developmental impact. A tentative categorisation of recipients is suggested in terms of capacity to use food aid either to create additional demand or in the context of food policies which use the resources made available by food aid to support agricultural development. There is a case for increasing food aid to almost triple current levels, largely going to low‐income countries, provided that both the recipients and donors commit themselves to policies which assure its constructive use.