The three-consecutive-day weighing method, in which foods as raw materials are weighed separately before they are cooked, is the most popular dietary assessment method among Japanese dietitians, because this dietary survey method has been adopted in the national nutrition survey for half a century. Under such circumstances, there have been only a few studies aiming to develop and validate a food frequency questionnaire (FFQ) method. The food list of a few FFQs ever used dealt with individual food items alone, but not mixed dishes. Therefore, a frequency questionnaire was very difficult for people not involved in cooking daily to answer. In order to enable subjects to answer a questionnaire more accurately, a new FFQ which has a food list with 122 single foods and recipes was developed. These foods and recipes were identified from 805 24hr-recalls of males and females aged 40-69 years old in Shiso County, Hyogo Prefecture, Japan. During consecutive 56 or 63 days, 67 junior college students in a dietitian course recorded their diet intake. Within a week after the diet record, they answered the newly developed FFQ. Pearson correlation coefficients between energy adjusted nutrient intakes assessed from the FFQ and those from dietary records were calculated. These correlation coefficients ranged from 0.21 for retinol potency to 0.74 for calcium. These data indicate that the newly developed FFQ provides a useful measurement of many nutrient intakes over a two-month period. J Epidemiol, 1996 ; 6 : S131-S136.