Underreporting of habitual food intake can be explained by underrecording and/or undereating. This study was designed to discriminate between the two errors mentioned, by measuring energy and water balance. Twenty-four lean female dieticians were recruited as subjects. Energy intake and water intake were measured for 1 wk with a weighed dietary record. Energy expenditure was estimated from measurements of resting metabolic rate, and measured physical activity with a triaxial accelerometer for movement registration. Water loss was estimated with deuterium-labeled water. Energy balance was determined by measuring the change in body mass over a nonrecording week (preceding the recording week) and over the recording week. Mean energy and water intake were 8.5 +/- 1.0 MJ/d and 2.3 +/- 0.5 L/d. The change in body mass in the nonrecording week was 0.1 +/- 0.6 kg and in the recording week -0.6 +/- 0.8 kg (paired t test; P = 0.02), indicating 16% undereating. Recorded water intake plus calculated metabolic water closely matched measured water loss (r = 0.93; P = 0.0001), which indicated a high recording precision. In conclusion, in the studied group of highly motivated lean women, there was 16% underreporting of habitual food intake, which could be explained by undereating.