The present study is a biochemical validation of a food-frequency questionnaire (FFQ) with optical reading, i.e. containing food portion photographs to help to assess quantities. Forty-four healthy subjects, non-smokers and not taking vitamin supplements, were recruited for the study. After completion of the questionnaire, subjects were asked to keep a 7 d weighed dietary record (7DR). Three 24 h urine samples were collected on 3 different days over the week of food recording for the analysis of urea-N, P and K. On the 4th day of food recording, blood was collected for determination of α-tocopherol, β-carotene and ascorbic acid. N, P and K determined in urines and from 7DR were significantly correlated (Spearman rank correlation test), r values being 0·77, 0·57 and 0·42 respectively. The correlations with the FFQ were significant only for N (r 0·45) and P (r 0·39). Blood ascorbic acid and β-carotene concentrations correlated with dietary intake when determined from 7DR (both r 0·44), but not when determined from FFQ. No correlation was found for α-tocopherol. The data obtained seem to prove the validity of the FFQ in defining eating patterns in terms of some nutrients, but not vitamins, at least as far asaq non-supplemented subjects are concerned. The way in which foods were grouped in the questionnaire could account for these results.