A food consumption survey was conducted in rural Bangladesh in January-March 1996 using a 24-h food weighing method. Intakes and adequacies of energy, vitamin A, vitamin C, iron and calcium were estimated for 81 children (aged 24-71 months) and 182 women (aged 18-45 years) using three sets of food composition tables. The local food composition table was modified with newer values for nutrient contents of foods identified as important for the contribution of vitamin A and iron. Significant differences in estimated dietary intakes and adequacies were found for vitamin A and iron when different sets of food composition values were used. This study shows that up-to-date chemical analyses of foods selected on the basis of a high frequency of consumption and a large contribution of the nutrients in focus is a useful approach in improving the quality of food composition tables, resulting in greater accuracy of estimates of nutrient intakes and adequacies.