The dietary intakes of 250 men and women aged from 17 to 60 years were evaluated by means of a self-administered questionnaire. The purpose of this study was to develop and optimise a self-administered questionnaire deliberately reduced to 20 questions. The results were compared with those of a dietetic interview using the diet history method. The correlation coefficients for the main nutrients were determined by multivariate regression analysis and were found to fall in the 0.44-0.80 range (P less than 0.001). It is concluded that the self-administered questionnaire is satisfactory and valid as compared with the diet history method. The degree of validity reached is such that efficient epidemiological studies can be conducted at small cost, and the subjects at risk due to the main dietary errors listed in the contemporary French population can be detected.