The aim of the present study was to describe the energy, nutrient and crude v. disaggregated food intake measured using 7 d diet diaries (7dDD) for the full baseline Norfolk cohort recruited for the European Prospective Investigation into Cancer (EPIC-Norfolk) study, with emphasis on methodological issues. The first data collection took place between 1993 and 1998 in Norfolk, East Anglia (UK). Of the 30 445 men and women, aged 40–79 years, registered with a general practitioner invited to participate in the study, 25 639 came for a health examination and were asked to complete a 7dDD. Data from diaries with data recorded for at least 1 d were obtained for 99 % members of the cohort; 10 354 (89·8 %) of the men and 12 779 (91·5 %) of the women completed the diet diaries for all 7 d. Mean energy intake (EI) was 9·44 (sd 2·22) MJ/d and 7·15 (sd 1·66) MJ/d, respectively. EI remained approximately stable across the days, but there was apparent under-reporting among the participants, especially among those with BMI >25 kg/m2. Micronutrient density was higher among women than among men. In conclusion, under-reporting is an issue, but not more so than that found in national surveys. How foods were grouped (crude or disaggregated) made a difference to the estimates obtained, and comparison of intakes showed wide limits of agreement. The choice of variables influences estimates obtained from the food group data; while this may not alter the ranking of individuals within studies, this issue may be relevant when comparing absolute food intakes between studies.