PURPOSE: To examine the effect of inappropriate initial antimicrobial therapy on the prognosis of patients with sepsis who were enrolled in a clinical trial of an immunomodulating agent conducted in 108 hospitals in North America and Europe. METHODS: We assessed initial antimicrobial choice and results of microbiologic cultures in 904 patients who had microbiologically confirmed severe sepsis or early septic shock. If a patient did not receive at least one antimicrobial agent to which the causative microorganisms were susceptible within 24 hours from the diagnosis of severe sepsis, then the initial antimicrobial treatment was considered to be inappropriate. A propensity score that adjusted for factors associated with inappropriate antimicrobial treatment was calculated and included in multivariable models to adjust for confounding. RESULTS: A total of 468 patients (52%) had documented bloodstream infection, and 211 patients (23%) received inappropriate initial antimicrobial therapy. Characteristics associated with inappropriate treatment were study enrollment in Europe, admission to surgery, nosocomial infection, infection with multiresistant microorganisms, and fungal or polymicrobial infection (all P <0.05). The 28-day mortality was 24% (168/693) for patients in the adequately treated group versus 39% (82/211) for patients receiving inappropriate initial antimicrobial therapy (P <0.001). After adjusting for comorbid conditions, severity of illness, site of infection, and the propensity score, inappropriate antimicrobial therapy was independently associated with increased mortality (odds ratio = 1.8; 95% confidence interval: 1.2 to 2.6). CONCLUSION: In a large cohort of patients with microbiologically confirmed severe sepsis, appropriate initial antimicrobial therapy was an important determinant of survival. New approaches aimed at improving detection and treatment of early sepsis are needed.