To evaluate the evidence that oxygen consumption (VO2) is pathologically dependent on oxygen delivery (DO2). Studies published since 1972 with their relevant bibliographies and computerized search of MEDLINE. All clinical papers reporting the relationship of: VO2 to DO2 in the adult respiratory distress syndrome (ARDS), sepsis, other critically ill patients, and normal individuals; cardiac output determined by measured VO2 to calculated VO2 from the arterial-mixed venous oxygen difference; blood lactate to DO2; and selected basic science studies. Study quality was assessed and all pertinent data were summarized. RESULTS OF DATA EXTRACTION: Normal individuals display physiologic dependence of VO2 at very low levels of DO2 (330 mL/min.m2). Pathologic dependence of VO2 on DO2 entails two concepts: a) VO2 varies directly with DO2 over a wide range of DO2 and b) of particular import, tissue oxygen extraction is compromised. This pathologic supply dependence was initially identified in patients with ARDS; subsequently, it has been demonstrated in patients with sepsis and in a variety of other critically ill individuals. There are substantial, but not uniform, data documenting this dependence of VO2 on DO2 in ARDS. In some studies, this relationship correlates best with increased lactate concentrations. However, increased blood lactate concentrations do not accurately track other evidence of tissue hypoxia. Some researchers have attributed the finding of this supply dependency to artifact, when VO2 is determined by the arterial-mixed venous oxygen difference. However, when these methods are compared, the correlation is excellent. Others have raised the concern that appreciable changes in VO2, even over short periods of time, may result in physiologic increases in DO2. However, when "control" groups have been contemporaneously compared with patients with ARDS using the same methodology, they have not shown supply dependency. Interwoven throughout the studies reviewed is overwhelming and uniform evidence that both mixed venous oxygen tension (PVO2) and mixed venous oxygen content (CVO2) correlate poorly with cardiac output, DO2, or VO2. The inconsistencies in identifying pathologic DO2 dependency may well reflect the unknown variables that exist in patients with ARDS, perhaps better labeled, multiple organ system failure. Pathologic dependence of VO2 on DO2, especially the inability to increase tissue oxygen extraction, is present in most patients with ARDS and many other critically ill individuals. PVO2 and CVO2 are both unreliable indicators of cardiac output, DO2, or VO2.