Because there are indications that eating habits may be an important factor in atherosclerosis, numerous surveys have been conducted in which quantitative estimates of the composition and caloric contents of diets were made. This study tested the accuracy of dietary survey methods by comparing calculated estimates of total calories and calories from fat with the actual number of calories, as measured by the bomb calorimeter. A remarkable inaccuracy in dietary calculations existed in all five types of sample collections made for the study but was greater in field samples than in hospital samples. Errors that can distort the dietary calculations include inaccuracy in the food-composition tables, in the data gathered from subjects in the field, or in the food samples provided, and errors in estimating weights or selecting tables. It is important, therefore, to check the data from dietary surveys with actual calorimetric determinations if quantitative statements are to be made on the correlation of dietary habits with disease.