Epidemiologic data on emergency department (ED) patients with sepsis are limited. Inpatient discharge records from 1979 to 2000 show that hospitalizations for sepsis are increasing. We examine the epidemiology of sepsis in US EDs and the hypothesis that sepsis visits are increasing. The National Hospital Ambulatory Medical Care Survey data (1992 to 2001) provided nationally representative estimates of frequency and disposition in adult ED visits for sepsis. Sepsis visits were identified using International Classification of Diseases, Ninth Revision, Clinical Modification codes; severe sepsis was defined as sepsis in conjunction with organ failure. Of 712 million adult visits during the 10-year period, approximately 2.8 million (0.40%, 95% confidence interval [CI] 0.33% to 0.46%) were related to sepsis. We found no significant increase in overall ED visits for sepsis from 1992 to 2001 (P for trend=.09). ED patients with sepsis were more likely to be elderly, non-Hispanic, and publicly insured and to arrive by ambulance compared with nonsepsis patients (all P<.01). The overall admission rate was 87% (95% CI 82% to 92%), with only 12% (95% CI 8% to 16%) of patients admitted to the ICU. The most frequent codiagnoses were pneumonia (13%), urinary tract infection (13%), and dehydration (11%). Severe sepsis accounted for 8% (95% CI 5% to 11%) of sepsis visits, for an annual incidence of 0.01%; 98% of patients with severe sepsis were admitted. In contrast to data from hospital discharges, ED visits for sepsis demonstrated no increase. Most ED visits for sepsis resulted in admission to non-critical care units.