This paper stresses the difficulty of resolving the many debates on food aid without closer attention to the impact of food aid programmes. It argues, however, that inadequate, casual, ex‐post evaluation must be replaced by a comprehensive impact assessment system. The data requirements of such a system are explored for different sorts of food aid activity and ways of simplifying the monitoring and evaluation task are discussed, based on selectivity, a step by step approach and the use of data‐economising methods. Specific proposals are made for an international initiative on impact assessment, with donor coordination and a strengthening of indigenous capability in recipient countries.