We compared Harris and Benedict [H & B; Harris, J. A., & Benedict, F. G. (1919). A biometric study of basal metabolism in man. Washington, DC: Carnegie Institution of Washington. p. 279.] predicted resting energy expenditure (REE) to values measured by indirect calorimetry in normal, uremic, diabetic, and uremic diabetic subjects. Predicted REE were overestimated (+9.2%, P<.005) in uremic subjects, and underestimated (-8.5%, P<.0001) in diabetic subjects. Uremic diabetic subjects were submitted to the opposite influences of diabetes and uremia on REE. Differences in body composition (lower fat-free mass in uremia and higher fat-free mass in diabetes) played a major role in these influences. In uremic diabetic subjects, predicted REE seemed well fitted to measured REE (biases <2%), but they were less correlated, and limits of agreement between predicted and measured REE were large. Although their mean REE seems normal, prediction by the H&B equation leads to important individual errors in uremic diabetic subjects: direct measurement of energy expenditure by indirect calorimetry may be helpful to precise the adequate energy content of a diet for these subjects.