Under field conditions, total daily energy expenditure (TDEE) has generally been estimated using time allocation techniques (the factorial method). However, recent work suggests that the factorial method underestimates TDEE relative to newer, more accurate methods such as doubly labelled water (DLW) and heart-rate (HR) monitoring. This study compares estimates of TDEE obtained by the factorial and HR-monitoring methods for a sample of 61 indigenous (Evenki; 17 males, 44 females) and 32 nonindigenous ("Russian"; 10 males, 22 females) subjects from three communities in Central Siberia. Energy expenditures obtained from the two methods were significantly correlated (r = 0.495; P < 0.0001), but the factorial method significantly underestimated TDEE relative to the HR-monitoring technique (8.95 +/- 2.73 vs. 8.25 +/- 1.34 MJ/d; P < 0.005). Interpopulational analyses of data compiled from this and other studies indicate that the factorial method consistently underestimates TDEE relative to DLW and HR monitoring and that the magnitude of underestimation increases with expenditure levels. Indeed, among sedentary populations, factorial estimates of TDEE converge on those of the other methods, whereas at high activity levels the disparity is quite large. These results imply that the daily energy requirements of many subsistence-level populations have been underestimated, thus providing an overly favorable picture of energy balance. Moreover, it is likely that underrepresentation of TDEE is most problematic in rural societies of the developing world which tend to have high activity levels and great risk of malnutrition.