The aim of the study was to assess the reproducibility and validity of the previous day food questionnaire (PDFQ) for schoolchildren. The questionnaire is illustrated with 21 foods and was designed for use at group level. The participants were 131 schoolchildren aged eight to ten years who were studying full-time in a public school in a city of Southern Brazil, in 2005. Reproducibility was assessed by applying the questionnaire twice on the same day. Validity was assessed by comparing the food items selected on the questionnaire and direct observations from three school meals on the previous day. The questionnaire presented high sensitivity, ranging from 73.4% (beans) to 95.5% (rice), and high specificity, ranging from 87.3% (fruits) to 98.8% (beans). It was concluded that the questionnaire was capable of generating reproducible and valid data for assessing the food intake of schoolchildren on the previous day.