Survey reports indicate that women perceive that their diets have changed in ways consistent with dietary guidelines recommended by national agencies. We have attempted to determine whether perceived change in food intake is a useful tool for estimation of either past intake or pattern of change in food consumption. Twin sisters of breast cancer patients, at obvious high risk of breast cancer, were aware of dietary guidelines concerning cancer, as indicated by beliefs about specific foods that should be increased or decreased to prevent cancer. Perception of change, as reported by them, was not found to be a reliable indicator of actual change in food frequency, as measured by the difference between sequential food frequency questionnaires. Consumption of high-fat foods was observed to have decreased over the interval, regardless of perceived change in consumption. Moreover, prediction of past food frequency on the basis of perceived change and current intake combined resulted in a less accurate appraisal of past diet than did the use of current intake alone. Perceived change in food frequency appears to be biased in different ways for different foods and seems to be influenced by beliefs about the role of diet in cancer. Studies of etiology should probably not rely on such methods.