Data from four central European dietary investigations carried out between 1982 and 1984 in the German Democratic Republic, Poland and the Federal Republic of Germany applying both a food frequency questionnaire (FFQ) and a quantitative recording instrument (24-h recall, 3-d record or 10- to 14-d record) were analysed to compare frequency of food consumption as reported by the FFQ with the corresponding information as derived from the quantitative instrument. The actual intake was found to be overestimated by the FFQ for categories of frequent consumption and underestimated for categories of rare consumption. Categories for which both instruments yielded similar estimates of frequency were found to vary between food items and to depend on the overall frequency of consumption of an item. This indicates that when using unvalidated FFQs and taking their FFQ-categories literally artificial differences of food consumption within the population(s) under study can be introduced. This implies also the tendency that the average food intake estimated through FFQs can yield unrealistically high values for items consumed frequently.