How we define sepsis has significant implications for clinical care, quality improvement, and regulatory policies. Current sepsis criteria identify heterogenous patients that vary widely in their clinical syndromes, triggering pathogens, and prognoses; one-third have viral or non-infectious processes and crude mortality rates vary 30-fold. Nonetheless, clinicians have been trained to treat all patients with possible sepsis immediately, aggressively, and uniformly with broad-spectrum antibiotics. Evidence continues to mount, however, that immediate antibiotics are only critical for patients with septic shock or multiorgan dysfunction while, patients with single organ dysfunction without shock can safely tolerate short delays until antibiotics. This allows time to clarify whether these patients are infected or not. We suggest modifying sepsis operational definitions to flag just those patients in whom short antibiotic delays are associated with increased mortality. This will help focus sepsis care where it is needed, aid antibiotic stewardship, and increase the validity of sepsis quality measures.