In winter 1988/89, 194 elderly adults (96 men and 98 women, aged 70-75 years) were interviewed, using the dietary history method. The relative validity of a dietary history method used in this elderly population was assessed. The record of intake of energy and nutrients from the dietary history method was compared with that obtained in the same group by means of a 3-day food record completed during the same period. For daily intake of energy and nutrients, the differences in mean intake were generally below 15%. But mean differences were significant for energy, fat, dietary fiber, and calcium. The two dietary methods correlated well for energy and nutrients. Pearson's correlation coefficient varied from 0.80 (dietary fiber) to 0.24 (vitamin A). The two methods classified about half of the subjects similarly into tertiles for energy and most nutrients. Gross misclassification occurred for 2-17% of the subject. Agreement between the two methods were poorest for vitamin A. The results suggest that the dietary history method may be inadequate for determining the exact level of mean and distribution of dietary intake in a group or the precise intake of an individual. But the method could be used to test differences in means between groups and to rank subjects as to their dietary intake within the study.