In Western countries underestimation of energy intake (EI) is considered a serious problem in dietary surveys, but information on the accuracy of EI among Japanese people is sparse. We compared estimated EI with measured total energy expenditure (TEE) in 21 Japanese women aged 20-22 y. Dietary intake was estimated by 7-d weighed dietary records (7dWR) and a self-administered diet history questionnaire (DHQ). TEE was measured for 25+/-3 d (mean+/-standard deviation, range: 16-27 d) by a uniaxial accelerometer. Both EI by 7dWR (1,498+/-305 kcal/d) and EI by DHQ (1,599+/-331 kcal/d) were significantly lower than TEE (1,865+/-179 kcal/d) (p<0.001 and p<0.01, respectively), but neither estimated EI nor the magnitude of EI underestimation (20+/-15% for 7dWR and 13+/-23% for DHQ) was significantly different between two methods (p=0.25 and p=0.22, respectively). The Spearman correlation (r) between TEE and EI was 0.51 (p=0.2) for 7dWR and 0.09 (p=0.71) for DHQ, indicating better ranking of individuals by 7dWR. The accuracy of EI (EI/TEE) was negatively associated with the percentage of EI from protein in 7dWR (r=-0.44, p=0.049) and positively associated with the percentage of EI from fat in both 7dWR (r=0.45, p=0.04) and DHQ (r=0.62, p<0.01), suggesting selective overestimation of protein and selective underestimation of fat. These results indicate not only underestimation of habitual EI but also selective under- and/or overestimation of macronutrients in both 7dWR and DHQ.