OBJECTIVES: To determine the incidence of systemic inflammatory response syndrome (SIRS), sepsis and severe sepsis in surgical ICU patients and define patient characteristics associated with their acquisition and outcome. DESIGN: One-month prospective study of critically ill patients with a 28 day in-hospital follow up. SETTING: Surgical intensive care unit (SICU) at a tertiary care institution. METHODS: All patients (n = 170) admitted to the SICU between April 1 and April 30, 1992 were prospectively followed for 28 days. Daily surveillance was performed by two dedicated, specifically-trained research nurses. Medical and nursing chart reviews were performed, and follow up information at six and twelve months was obtained. RESULTS: The in-hospital surveillance represented 2246 patient-days, including 658 ICU patient-days. Overall, 158 patients (93%) had SIRS for an incidence of 542 episodes/1000 patients-days. The incidence of SIRS in the ICU was even higher (840 episodes/1000 patients-days). A total of 83 patients (49%) had sepsis; among them 28 developed severe sepsis. Importantly, 13 patients had severe sepsis after discharge from the ICU. Patient groups were comparable with respect to age, sex ratio, and type of surgery performed. Apache II score on admission to the ICU and ASA score at time of surgery were significantly higher (p < 0.05) only for patients who subsequently developed severe sepsis. The crude mortality at 28 days was 8.2% (14/170); it markedly differed among patient groups: 6% for those with SIRS vs. 35% for patients with severe sepsis. Patients with sepsis and severe sepsis had a longer mean length of ICU stay (2.1 +/- 0.2 and 7.5 +/- 1.5, respectively) than those with SIRS (1.45 +/- 0.1) or control patients (1.16 +/- 0.1). Total length of hospital stay also markedly differed among groups (35 +/- 9 (severe sepsis), 24 +/- 2 (sepsis), 11 +/- 0.8 (SIRS), and 9 +/- 0.1 (controls, respectively). CONCLUSIONS: Almost everyone in the SICU had SIRS. Therefore, because of its poor specificity, SIRS was not helpful predicting severe sepsis and septic shock. Patients who developed sepsis or severe sepsis had higher crude mortality and length of stay than those who did not. Studies designed to identify those who develop complications of SIRS would be very useful.