This study was conducted during the preparation of a case-control study on patients with adenomatous polyps or cancer of the large bowel. It was done to compare two dietary history methods for assessing individual current dietary intakes. Subjects were interviewed concerning their food consumption by meal in one of the questionnaires and by broad food groups in the other questionnaire. Two groups of 20 volunteers, comparable according to sex and age, were interviewed by a dietitian who used one of the questionnaires. Data on diets obtained with the questionnaire were compared with those of a 14-day dietary record. The subjects were informed of the importance of the validation procedure. Whatever the type of questionnaire, mean daily intakes of nutrients and foods were rather similar for the questionnaires and for the 14-day records. Thus, there were more significant correlations between the questionnaire by meal and the record than between the questionnaire by food group and the record. When using tertiles, it appears that the questionnaire by meal was better at classifying individuals with regard to their food intake than the questionnaire by food groups. Although extrapolating these results to a patient population is not straightforward, it seemed likely that interviewing patients on diet without using a pattern of meals would prove even less reliable than for healthy subjects. The questionnaire by meal was preferred to the questionnaire by food groups for the ongoing case-control study. Further studies are needed to know whether these results could be extrapolated to studies on past diet and to non-Latin populations.