The optimal duration for courses of antibiotic therapy following emergency abdominal surgery was examined. The length of postoperative administration was based on the operative findings of contamination versus infection and the degree of the latter. A total of 163 patients (mean APACHE II score 7) were stratified into four groups: group 1 (60 patients), no postoperative antibiotics; group 2 (32), antibiotic therapy for 24 h; group 3 (48), administration for 48 h; and group 4 (23), antibiotic therapy for 72 h to 5 days. Three patients (2 per cent) died. Wound infection developed in 12 patients (7 per cent) and postoperative intra-abdominal infection in two (1 per cent). Antibiotics were stopped according to the protocol in 28 patients in spite of continued fever; one developed a subhepatic abscess and three had wound infections. Distinguishing contamination from infection and operative stratification of the latter allowed a successful 'minimal' postoperative antibiotic policy to be employed.