Chemical analysis of food actually eaten would be the most accurate method for analyzing the diets of children, but this technique is not feasible for most nutrition staffs. An improved 24-hr. dietary recall can be used by a small, well trained staff to collect more reliable data on a large number of school children. Vigilant monitoring of school lunch operations, incorporation of known recipes in the ETNV, and organizing probing techniques are necessary to insure the reliability of the tool. The low coefficients of variation of duplicate recalls noted in the study indicate that the error of measurement between interviewers is small, if the tool is carefully tested and sophisticated before use in the field and if observers are carefully trained by a written protocol.