The turnover rates of water and carbon dioxide have been estimated at frequent intervals after a single dose of 2H2 18O had been given to an adult monitored continuously in a whole body calorimeter for 5 d. The integrated CO2 production rate for 5 d, estimated from the differential rates of 2H and 18O excretion in urine, was within 2 per cent of the actual CO2 output, based on an improved method of calculation. Changes occurred in the abundance of 2H and 18O in response to activity, so that the timing of samples was important. It was found better to compute the production rate of CO2 from multiple samples rather than from measurements at only two points of time. The proportion of total water lost by evaporation proved to be less than previously assumed. Direct validation of the technique requires calorimetric studies of 2 to 3 weeks duration, in either animals or human subjects. The potential value of this method for measurement of energy expenditure in man is such that exacting tests of its accuracy and precision are needed before it is adopted for general use in metabolic and epidemiological studies.