A prospective randomized trial has compared a broad spectrum ureidopenicillin with a broad spectrum cephalosporin for prophylaxis against the aerobic organisms encountered during elective colonic surgery. Even though only two doses of antibiotics were administered the incidence of severe sepsis was low. Severe wound infection occurred in three of the patients receiving mezlocillin and metronidazole (6 per cent) compared with six in the group receiving cefuroxime and metronidazole (13 per cent). Minor wound sepsis was recorded in 24 per cent of patients receiving mezlocillin and metronidazole compared with only 11 per cent after cefuroxime and metronidazole. There were two episodes of septicaemia, one in each group, and three abscesses, all of which occurred in patients receiving metronidazole and mezlocillin. The total number of surgically related infections was, however, significantly less with cefuroxime and metronidazole (N = 13) compared with mezlocillin and metronidazole (N = 23; P less than 0.03). Escherichia coli was the principal organism responsible for surgically-related postoperative sepsis: (22 isolates: 14 mezlocillin and eight cefuroxime) all of which sensitive to the agents used. Pseudomonas aeruginosa was recovered from 10 patients (three mezlocillin and seven cefuroxime), all of the isolates were resistant to both antibiotics and were associated with severe morbidity. There were 11 isolates of Staphylococcus spp. (nine mezlocillin and two cefuroxime: P less than 0.03). Postoperative diarrhoea occurred in six patients, all were in the group receiving cefuroxime and metronidazole. (Clostridium difficile was recovered from the stool in three of which one was associated with Cl. difficile cytotoxin.)