Three-day, rather than seven-day, food records are frequently used because mis-reporting of food intake is believed to increase with recording period. Data from the National Diet and Nutrition Survey of adults were used to explore trends in reported energy intake (REI) with day of recording, and to compare average REIs using the common method of three consecutive days (Thursday to Saturday) to the complete seven-day record. Although REIs decreased from days 2 to 7, this was by a quantitatively insignificant 49kJ per day (P = 0.026), and well within the measurement error of recorded food intakes. Furthermore, REIs were lowest on the first recording day. The three-day was greater than the seven-day average (mean 0.30 (SD 1.0) MJ, range 3.3 to +4.7MJ). Those who reported higher than average energy intakes did so disproportionally at the weekend. These results support the use of longer rather than shorter reporting periods.