Sources of error in the interpretation of respiratory data are evaluated and reviewed with special reference to the detailed composition of foods. Estimates of fuel utilization or synthesis are 12-fold more sensitive to errors in the nonprotein respiratory quotient than is the heat equivalent of oxygen. Estimates of protein oxidation from nitrogen excretion can be in error from +14 to -39% of the true value. Heat equivalents of oxygen, respiratory quotients, and urinary nitrogen-to-oxygen conversion ratios are considered for 60 artificial and 101 conventional food proteins, 36 artificial and 125 conventional food fats, and the different carbohydrates contained in these foods. It is concluded that there is considerable uncertainty when the mix of fuels utilized is assessed accurately. Accuracy is best within 5% of the true values. This analysis is completed with descriptions of some physiological sources of error in an appendix.