Of 330 consecutive patients with liver trauma having a celiotomy over a 5-year period, 295 (89%) survived more than 72 hours. Of these 295, 35 (12%) developed sepsis, and 11 (31%) of these septic patients died. The sources of the sepsis in 30 of these patients included: abdominal abscesses--23, pneumonia or empyema--seven, acalculous cholecystitis--two, gangrene of right colon--two, and thigh abscess--one. In five other patients, the source of the sepsis was not found, even at autopsy. The mortality rate in the 30 patients with one or more identifiable foci of infection was 23%. In contrast, when the source of the sepsis could not be found, the mortality rate was 80% (4/5) (p less than 0.05). Factors associated with an increased incidence of abdominal abscess included: splenectomy, 75% (3/4); liver packs, 63% (5/8); 20+ units of blood, 57% (8/14); Class IV-V liver injury, 35% (8/23); 10-19 units of blood, 25% (7/28); colon injury, 19% (7/36); and open (Penrose) drainage of the abdomen, 11% (23/213). None of 82 patients without drains developed an intra-abdominal abscess. Thus early control of an identifiable source of infection provides the best results with sepsis following liver trauma. The most effective method for preventing intra-abdominal abscesses appears to be avoidance of drains in mild (Class I-II) liver injuries. The use of a closed system in the most severe injuries is still controversial and needs to be addressed in a prospective trial.