Objective To investigate whether oxygen consumption (o2) is dependent on oxygen delivery (Ḋo2) in adult respiratory distress syndrome (ARDS) and non-ARDS acute respiratory failure. Design Intervention study of a consecutive sample of patients admitted to the ICU with the diagnosis of acute respiratory failure. Setting Tertiary care center. Patients Thirteen consecutive patients with a diagnosis of ARDS and 11 with a diagnosis of respiratory failure not due to ARDS. Patients were monitored with an oximetric pulmonary artery catheter and mechanically ventilated. Interventions Ḋo2 was decreased by the application of positive end-expiratory pressure (PEEP) (20 cm H2O), and subsequently increased by an iv infusion of dobutamine (10 μg/kg-min). Results After the application of PEEP, Ḋo2 decreased significantly in both groups. However, o2 decreased significantly (p < .01) only in the ARDS group. When dobutamine was infused, Ḋo2 increased significantly (p < .01) in both groups, but o2 increased only in ARDS patients. Ḋo2 correlated significantly with o2 both in ARDS (r2 = .81, p < .01) and in non-ARDS (r2 = .38, p < .05) patients. The correlation coefficient was significantly higher for ARDS than for non-ARDS patients. Comparing the slopes of the regression lines, a stronger dependency of o2 on Ḋo2 was found in ARDS than in non-ARDS respiratory failure (p < .001). The oxygen extraction ratio correlated with Ḋo2 in non-ARDS patients (r2 = .49, p < .05), but not in ARDS patients. Conclusions o2 is dependent on Ḋo2 over a wide range of Ḋo2 values in acute respiratory failure. This dependency phenomenon is much stronger in ARDS than in respiratory failure due to the causes. Due to the abnormal dependency of o2 on Ḋo2, changes in the oxygenation status may not be reflected by changes in mixed venous oxygen saturation in ARDS.