Energy balance can be estimated in tissues, body segments, individual subjects (the focus of the present article), groups of subjects and even societies. Changes in body composition in individual subjects can be translated into changes in the energy content of the body, but this method is limited by the precision of the techniques. The precision for measuring fat and fat-free mass can be as low as 0.5 kg when certain reference techniques are used (hydrodensitometry, air-displacement plethysmography, dual-energy X-ray absorptiometry), and approximately 0.7 kg for changes between two time points. Techniques associated with a measurement error of 0.7 kg for changes in fat and fat-free mass (approximately 18 MJ) are of little or no value for calculating energy balance over short periods of time, but they may be of some value over long periods of time (18 MJ over 1 year corresponds to an average daily energy balance of 70 kJ, which is < 1% of the normal dietary energy intake). Body composition measurements can also be useful in calculating changes in energy balance when the changes in body weight and composition are large, e.g. > 5-10 kg. The same principles can be applied to the assessment of energy balance in body segments using dual-energy X-ray absorptiometry. Energy balance can be obtained over periods as short as a few minutes, e.g. during measurements of BMR. The variability in BMR between individuals of similar age, weight and height and gender is about 7-9%, most of which is of biological origin rather than measurement error, which is about 2%. Measurement of total energy expenditure during starvation (no energy intake) can also be used to estimate energy balance in a whole-body calorimeter, in patients in intensive care units being artificially ventilated and by tracer techniques. The precision of these techniques varies from 1 to 10%. Establishing energy balance by measuring the discrepancy between energy intake and expenditure has to take into consideration the combined validity and reliability of both components. The measurement error for dietary intake may be as low as 2-3% in carefully controlled environments, in which subjects are provided only with certain food items and bomb calorimetry can be undertaken on duplicate samples of the diet. Reliable results can also be obtained in hospitalised patients receiving enteral tube feeding or parenteral nutrition as the only source of nutrition. Unreliability increases to an unknown extent in free-living subjects eating a mixed and varied diet; thus, improved methodology is needed for the study of energy balance.