Progress in the care of the critically ill patient with life-threatening infection has been hampered by inconsistent, often confusing terminology. The clinical syndrome of sepsis-familiar to all yet definable by none-describes a highly heterogeneous group of disorders with different causes and differing prognoses. The imminent availability of mediator-directed therapy has created a sense of urgency to develop better methods for delineating discrete clinical syndromes and to modulate the host response, which may bring both benefit and harm, depending on the clinical circumstances. The term systemic inflammatory response syndrome (SIRS) was introduced several years ago to describe the familiar clinical syndrome of sepsis, independent of its cause. SIRS can result from trauma, pancreatitis, drug reactions, autoimmune disease, and a host of other disorders; when it arises in response to infection, sepsis is said to be present. SIRS describes a dynamic process that has adaptive survival value for the host. The maladaptive consequence of this process in the critically ill patient is the development of progressive but potentially reversible remote organ dysfunction-the multiple organ dysfunction syndrome. The development of cogent conceptual frameworks for classification of the septic response in critically ill patients is more than a question of linguistic pedantry. Optimal therapy presupposes identification of an homogeneous patient population with a characteristic disease process and a predictable response to an intervention. Although progress has been made in identifying such groups of critically ill patients, the disappointing results of clinical trials of agents that so clearly demonstrate efficacy in animal models indicates that considerable work remains.