OBJECTIVE To determine whether continuous gas flow in the breathing circuit or an airleak around the tracheal tube cuff will introduce errors into the measurement of oxygen consumption (VO2) with indirect calorimetry. METHODS Nonrandomized, controlled trial. METHODS Experimental laboratory. METHODS Ten healthy, anesthetized mongrel dogs, weighing 8 to 12 kg. METHODS Data were recorded at seven levels of flow, from 0 to 12 L/min in excess of minute ventilation, through a continuous breathing circuit. Data were recorded at five levels of tracheal tube cuff leak from 0% to 40% of inspiratory minute volume. RESULTS VO2 was measured using an indirect calorimeter with constant internal gas flow and calculated from results of blood gas analysis, cooximetry, and thermodilution cardiac output determinations at all levels of continuous breathing circuit flow and cuff leak. BP, heart rate, respiratory rate, arterial and mixed venous blood gases, and body temperature were measured to assess stability of cardiopulmonary function. Continuous breathing circuit flow did not affect the accuracy of indirect calorimetry until the total flow reached a critical value (11.5 L/min) that was slightly below the internal flow constant of the metabolic monitor (12.4 L/min). At higher circuit flows, measured VO2 decreased in a linear fashion, while calculated VO2 remained unchanged. Above the critical flow, the error of indirect calorimetry correlated significantly only with the total circuit flow (r2 = .64), not with the exhaled concentration of CO2 (r2 = .005) or the inspiratory-expiratory oxygen difference (r2 = .004). The continuous flow rate at the critical circuit flow was 66 +/- 15% of the subjects' peak inspiratory flow. Increasing tracheal tube cuff leak produced a progressive decrease in measured VO2 but not in calculated VO2. The difference between measured and calculated VO2 was linearly related to the magnitude of the leak (r2 = .56), and was statistically significantly larger at all levels of cuff leak, when compared with measurements during complete cuff seal. CONCLUSIONS An indirect calorimeter in which measurement of VO2 is based on internal constant flow rather than spirometry can be used to accurately measure VO2 from a continuous-flow breathing circuit, if the total circuit flow is less than the internal flow. This limitation may restrict the use of continuous flow to a level below the subject's peak inspiratory flow. The accuracy of indirect calorimetry cannot be guaranteed for any amount of tracheal tube cuff leak.