The aim of the study was to investigate the accuracy of children’s reports of their consumption of a school meal. Nutritional intake from the school lunch was studied for 56 healthy 8-year-olds and 43 13-year-olds. Information as to nutritional intake was obtained by personal interviews within the limits of 24-hour recall. The results were compared with those of the double-portion technique, which involves weighing and chemical analysis of a double portion, with subtraction of left-overs. Energy, protein, fat and iron content were determined. The data for the individual meals were spread over a wide range. On a group basis, however, the differences between nutrient data obtained by recall and by analysis were levelled out. In spite of a number of possible sources of error in both the double-portion method and the recall, the results obtained with the two methods gave median values for amounts consumed and for nutritional data that were in fairly good agreement. The median values for energy and nutrient content of a given double portion obtained by calculation from food composition tables and by chemical analysis were in good agreement. It is concluded that the groups of 8- and 13-year-old children in this investigation were able to provide an acceptable picture of their consumption of a single meal in a dietary recall covering the past 24 h.