Our objectives were to estimate the degree of misreporting energy intake (EI) and analyze associations with previous BMI, current BMI, or both. The study was part of the Adiposity and Genetics Study follow-up study including 309 Danish men (age 40-65 y) originally sampled from the obligatory draft board examination. Height and weight were measured at the mean ages of 20 (draft board), 33, 44, and 49 y (current age). Obesity was categorized as BMI >or= 31 kg/m(2). Dietary intake for 7 d and physical activity (PA) level (PAL) were self-reported. Resting metabolic rate (RMR) was measured in a ventilated hood system. By comparing EI with energy expenditure and assuming energy balance, reporting accuracy (RA) was estimated as EI/(RMR.PAL). A plausibility interval was calculated to encompass specific variation components of EI, RMR, and PAL; the specific 95% plausibility interval was 1.00 +/- 0.35. Participants were categorized as underreporters (RA <or= 0.65), plausible reporters (0.65 < RA <or= 1.35), or overreporters (RA > 1.35) of EI. The relation between RA and BMI was studied through linear regression analysis. Overall, the RA was (mean +/- SE) 0.76 +/- 0.01. Of 309 participants, 35% underreported and 7% overreported. Whether stratified for current BMI or draft board BMI, the obese men were more likely to underreport than those who were not obese. Among those currently not obese, underreporting was more prevalent among those who were obese at the draft board examination (44%) than among those who were not (21%). Regression analysis showed that both previous and current BMI and their combination were significantly associated with RA. Thus, underreporting of dietary intake seems to be associated with not only current BMI but also with current BMI in combination with previous BMI.