OBJECTIVE To assess the quality of a narrative form diet history (DH). DESIGN Reproducibility assessed with data obtained at 6-week intervals. Criterion validity assessed using energy intake to estimated energy expenditure (EI:BMR) cutoff limits. Relative validity assessed by comparing results for energy and macronutrients at baseline and month 2 of an intervention trial with those obtained from 3-day food records (FR). SETTING Community-based dietary intervention trials for the study of metabolic syndrome in the Illawarra region of New South Wales, Australia. SUBJECTS Reproducibility: 43 healthy female volunteers. Mean age 58.72 years (range 50-67), mean body mass index (BMI) 25.79 (range 21-36). VALIDITY 45 healthy volunteers--18 males (mean age 46.9 years, mean BMI 27.8), and 27 females (mean age 45.7 years, mean BMI 26.2), attending a study on the effect of diet on metabolic variables. RESULTS Reproducibility: wide SD values indicated a high degree of intraindividual variation, but correlation coefficients were comparable to those of similar studies. VALIDITY underreporting was inconsistent with individuals, but was greater with the DH than the FR at each stage of the dietary trial (significant at month 1, P < 0.01). Underreporters indicated higher intakes of protein during the trial (P < 0.05). Weaker associations were found between the DH and FR data for energy values, but there was strong agreement for per cent fat consumed as saturated and monounsaturated fat at each stage of the trial. CONCLUSIONS The narrative form DH performs as well as standardized interviews, but more work needs to be done at the micro level, focusing on aspects which deal with foods likely to be underreported in the particular study context. This can be achieved through continued research using combined methodologies.