Accurate assessment of the habitual dietary intake is pre-requisite to accurate studies on diet and health, but underestimation of the dietary intake has been apparent in numerous studies. If the underestimation of dietary intake was consistent, a solution to the problem would be relatively easy, because such techniques as energy adjustment should improve estimates of the food and nutrient intake. If, however, underestimation of the dietary intake was a selective phenomenon, it would be much more difficult to solve the problem. A limited number of studies in Western countries have examined whether all foods and nutrients are underestimated to the same degree or only specific foods and/or nutrients are selectively underestimated by using the observed food intake, the measured total energy expenditure with/without 24-hour urinary nitrogen excretion, and the Goldberg cut-off technique as reference methods. These investigations are summarized in the present review. The bulk of the data from these studies suggests that when the energy intakee has been underestimated, such underestimation of the food and nutrient intake is selective rather than consistent. However, little information about those foods that are selectively underestimated is available, so further research is needed to identify such foods.