The primary function of the heart and lungs is to generate a flow of oxygenated blood to respiring tissues to sustain aerobic metabolism. Teleologically, such a transport system has several basic requirements. It should be energy efficient, avoiding unnecessary cardiorespiratory work, but it should be sensitive to the fluctuating demands of cellular metabolism. Ideally, metabolic demand and oxygen distribution should be matched regionally when at rest, during exercise, and in different disease states. Finally, oxygen should pass efficiently across the extravascular tissue matrix. The mechanisms that control oxygen distribution are complex and not completely understood. In the critically ill patient, these mechanisms may have an important role in determining the clinical outcome. The relationship between oxygen delivery and consumption has not been clearly established despite considerable investigation during the last decade. However, these variables are often measured to define a population of critically ill patients in whom oxygen consumption is limited by oxygen delivery, the state of so-called delivery-dependent oxygen consumption or pathologic supply dependency. The recent literature in critical care and many leading intensive care units has emphasized the importance of raising oxygen delivery to "supranormal" levels in an attempt to satisfy the increased metabolic demands of these patients. This practice has been justified by the observation that increasing oxygen delivery improves oxygen debt and outcome in postoperative surgical patients requiring intensive care. In the severely hypovolemic patient, most physicians would agree that volume replacement to improve oxygen delivery must be beneficial. However, in patients with more complex problems, including sepsis, cardiovascular collapse, and hypoxic hypoxemia, controlled trials to examine the influence of such strategies on clinical outcome have produced conflicting data. Several methodologic factors may have contributed to these contradictory and often controversial results. These factors include failure to define the disease and patient population adequately, the relationship between the time of investigation and the evolution of the disease process, and the accuracy and frequency of measurement.(ABSTRACT TRUNCATED AT 400 WORDS)