Because of large stores of CO2 in different body tissues, metabolic change cannot be detected by measuring gas exchange until the CO2 stores have adapted to the new situation. Similarly, changes in the CO2 stores not due to metabolic alterations, may lead to error in gas exchange measurements. We studied CO2 production (VCO2) and oxygen consumption (VO2) in mechanically ventilated ICU patients, where CO2 stores were altered by: a) changing minute ventilation by 15%, b) reducing body temperature, and c) changing the level of sedation. Expired gases went through a mixing chamber and were analyzed continuously by a mass spectrometer. Signals from this instrument, together with gas-volume signals from the ventilator, were fed to a computer for calculation of VO2 and VCO2. Twenty to 120 min were required to reach a stable level, depending on the patient's size and circulatory response. Similar results were obtained by computer simulation using a five-compartment model of CO2 stores. These experiments indicate that measuring VO2 (for calculation of metabolic respiratory quotient [RQ]) in ventilated patients should occur after the patients maintain a 60-min period of stable body temperature and awareness. Ventilatory variables should not be changed substantially during the 90-min period before gas sampling. Cardiac output and muscle blood flow should not have changed 2 to 3 h before measuring RQ. If muscle blood flow is low, the stable periods for body temperature and ventilatory variables should be increased.