A study was intended to estimate energy intake (EI) for a nutrition intervention study. Subjects were 17 elderly men (age: 72 +/- 5 years) and 11 elderly women (age: 67 +/- 4 years). Two methods were used to measure EI: a 4-day dietary record (DR group; n = 12) or a dietary questionnaire (DQ group; n = 16). Subjects were fed for 3 weeks according to this intake during an intervention period, and energy expenditure (EE) was measured with doubly labelled water to verify the resulting figure for EI. Body weight and body composition were measured at the beginning of the second week and at the end of the third week, and metabolizable energy (ME) was calculated as gross energy intake (GEI) minus energy in faeces and urine. GEI, ME and EE did not differ between the DR and DQ groups (mean +/- SD for GEI: 10.09 +/- 1.21 and 9.29 +/- 1.36; ME: 9.09 +/- 1.28 and 8.34 +/- 1.31; EE: 10.13 +/- 1.57 and 9.25 +/- 0.35 MJ/d). In both groups ME was significantly lower than EE (DR group P < 0.05; DQ group P < 0.01). Body weight decreased significantly during the intervention period (mean +/- SD for DR group: -0.64 +/- 0.50 kg, P < 0.001; DQ group: -0.86 +/- 0.90 kg, P < 0.01). The change in body weight was significantly correlated with energy balance (ME - EE; P < 0.05). As shown from the results of the body composition measurements, body weight loss was a decrease of fat mass.(ABSTRACT TRUNCATED AT 250 WORDS)