In this article, we review some fundamentals of indirect calorimetry in mice and rats, and open the discussion on several debated aspects of the configuration and tuning of indirect calorimeters. On the particularly contested issue of adjustment of energy expenditure values for body size and body composition, we discuss several of the most used methods and their results when tested on a previously published set of data. We conclude that neither body weight (BW), exponents of BW, nor lean body mass (LBM) are sufficient. The best method involves fitting both LBM and fat mass (FM) as independent variables; for low sample sizes, the model LBM + 0.2 FM can be very effective. We also question the common calorimetry design that consists of measuring respiratory exchanges under free-feeding conditions in several cages simultaneously. This imposes large intervals between measures, and generally limits data analysis to mean 24 h or day-night values of energy expenditure. These are then generally compared with energy intake. However, we consider that, among other limitations, the measurements of Vo(2), Vco(2), and food intake are not precise enough to allow calculation of energy balance in the small 2-5% range that can induce significant long-term alterations of energy balance. In contrast, we suggest that it is necessary to work under conditions in which temperature is set at thermoneutrality, food intake totally controlled, activity precisely measured, and data acquisition performed at very high frequency to give access to the part of the respiratory exchanges that are due to activity. In these conditions, it is possible to quantify basal energy expenditure, energy expenditure associated with muscular work, and response to feeding or to any other metabolic challenge. This reveals defects in the control of energy metabolism that cannot be observed from measurements of total energy expenditure in free feeding individuals.