This first article of the supplement describes the rationale for the Department of Veterans Affairs Cooperative Study, Processes, Structures, and Outcomes in Cardiac Surgery, which was designed to demonstrate statistically and clinically meaningful linkages between processes and structures of care and the outcomes of that care. United States health care is in an era of great enthusiasm for the use of health care outcomes to assess and improve quality of care. An important reason for this enthusiasm is the concern that processes and structures of care, which traditionally have been selected arbitrarily without valid linkages to favorable outcomes, may not result in the desired outcomes of care. Furthermore, health care outcomes are intrinsic to the definition of quality of care and should be relatively free of preconceived biases about how care should be provided. However, the limitations to outcomes-directed quality improvement have been inadequately recognized. These limitations include the following: (1) mortality, the most commonly used outcome, is usually sufficiently rare, resulting in inadequate statistical power; (2) nonfatal outcomes are much more difficult to measure reliably; (3) outcomes may not be measurable for an extended period of time after the care episode, making linkage to quality improvement inefficient; and (4) patients often desire good processes of care as well as favorable outcomes. A review of the literature found relatively few reports linking processes and structures of care to favorable outcomes. Significant relationships between processes of care and outcomes have been reported for several medical conditions (congestive heart failure, acute myocardial infarction, pneumonia, and stroke) when the patient has been considered the unit of analysis. However, there is a paucity of published meaningful process-outcome or structure-outcome linkages for surgical conditions or for any conditions when the hospital has been the focus of analysis. The authors concluded that quality improvement will proceed most efficiently and effectively if all three elements of Donabedian's quality triad (processes, structures, and outcomes) are used and if the processes and structures chosen have been demonstrated to be associated with desired outcomes of care.