To create a suitable instrument to estimate intakes of total calories, protein, carbohydrate, fats (saturated, mono and polyunsaturated), alcohol, cholesterol, fibre, vitamin A and vitamin C in epidemiological studies conducted in Spain, a food frequency questionnaire was developed and tested. In particular, the questionnaire was designed to be used in a large population-based case-control study of dietary factors in relation to breast and colorectal cancer among women from different Spanish regions. After identifying the most important food sources of the relevant nutrients in the study population, the final version of the questionnaire asked about consumption of 118 food items. Its reproducibility and validity were tested among 147 Spanish women aged 18-74 years. These subjects were asked to complete the questionnaire before and after completing four 4-day food records. The records were obtained at 3-month intervals designed to represent daily and seasonal changes (between 1990 and 1991). Using the information available from standard Spanish food composition tables, an ad hoc computer program was created to translate food consumption into nutrient intake. The reproducibility of the questionnaire was assessed by means of estimating correlations between nutrient scores measured with the same instrument twice, with a period of 1 year between estimates. Pearson correlation coefficients ranged from 0.51 for saturated fat to 0.88 for alcohol. In the validity study, correlation coefficients between diet records and the first and second questionnaires ranged between r = 0.20 for vitamin A and r = 0.88 for alcohol.(ABSTRACT TRUNCATED AT 250 WORDS)