A related issue is whether septic shock identified within 24 h of ICU admission might play out differently than shock that develops later in the ICU stay. In an observational study of 65 patients, Roman-Marchant and colleagues (see page 173) provide evidence that freshly admitted septic patients have greater degrees of physiologic derangement, resulting in higher severity-of-illness scores, yet better outcomes than patients already in the ICU when sepsis occurs. Differences in infecting organisms and changes in immunomodulation over time are offered as potential explanations. There may be other explanations to explore: altered catecholamine responsiveness, progressive capillary “leak,” inability to supply adequate nutrition, unmeasured endocrine changes, and adverse effects of interventions such as pulmonary artery catheterization or mechanical ventilation. The simplest explanation, though, may be that the early and late sepsis populations are qualitatively quite different. The mere fact of being in the critical care unit implies failure to respond to prior therapy, and perhaps self-selection by nonresponders contributes to the observation of poor outcome with late sepsis. This article is an important first step in refining risk stratification, and raises the question of whether other, objective markers could better delineate the “early” from the “late” sepsis population. Genetic polymorphism, for example, has a role in determining endothelial6and adrenergic7function. As genetic markers of susceptibility to sepsis are identified,8 it will be intriguing to see if there are markers for “therapeutic resistance,” and if the late-sepsis population carries an identifiable genetic profile. For now, we have just epidemiologic evidence that the populations differ, and that timing of sepsis onset must be considered in interpreting clinical trials and when judging quality of health care based on clinical outcomes.