An accurate, simple method for assessing energy expenditure in individuals and in free-living populations continues to be elusive. To compare estimates of energy expenditure (EE) from a combination of two previously validated physical activity questionnaires: Tecumseh Occupational (EE(TEC)) and a 4-wk history version of the Minnesota Leisure Time Physical Activity that included household activities (EE(MNLTPA)) and EE from sleep (EE(SLEEP)), to EE obtained from doubly labeled water (EE(DLW)). We studied free-living males (n = 24) eating a controlled diet designed to maintain body weight and determined EE from doubly labeled water (DLW) during 14 days and EE from physical activity instruments used in epidemiological studies (EE(TEC) and EE(MNLTPA)). There was excellent agreement between EE(DLW) (mean +/- SEM, 13.55 +/- 0.38 MJ/d) and EE(TEC) + EE(MNLTPA) + EE(SLEEP) (EE(TOTAL1)) (13.79 +/- 0.89 MJ/d) with a difference of only 1.0% +/- 5.4%. When the EE from watching TV, reading, and childcare activities was added the total EE (EE(TOTAL2)) (14.87 +/- 0.90 MJ/D) overestimated EE(DLW) by 8.9% +/- 5.4%. Both of these estimates of EE had significant regressions against EE(DLW) (EE(TEC) + EE(MNLTPA) + EE(SLEEP), R(2) = 0.38, P < 0.001; EE(TOTAL2), R(2) = 0.39, P < 0.001). Men whose occupations involved significant intermittent moderate activity had the largest disagreement between EE(DLW) and estimates from the questionnaires. This investigation demonstrates that a combination of previously validated physical activity questionnaires can be used to accurately determine the mean energy expenditure of a population of employed males.