When prophylactic antibiotics are used in abdominal surgery it is customary to give the first dose before the operation. Whilst intra-operative antibiotics may be effective in elective surgery, there may be an advantage to starting pre-operatively when there is already an infective focus such as appendicitis. Antibiotics started pre-operatively (group P) have been compared with antibiotics started after initial abdominal exploration (group T). Three intravenous doses of 500 mg metronidazole plus 1 g cephazolin were given in a randomized, double-blind study of 700 emergency and elective high-risk abdominal operations. Antibiotic plasma concentrations at the end of the operation were significantly lower in group P but lay well within the therapeutic range. Wound infection rates, which included minor and delayed infections, were similar in both groups (group P, 57 of 342, 16·7 per cent; group T, 55 of 358, 15·4 per cent; 95 per cent confidence intervals for the difference being −4·1 to +6·7 per cent. In appendicitis, wound infection rates were 12·1 and 13·9 per cent for groups P and T respectively. However, non-fatal deep sepsis was more common in group P (nine cases) than in group T (two cases) (χ2 = 4·9, P < 0·05). Postoperative infection was twice as common in obese patients whose body mass index (BMI) was ⩾ 26 (39 of 132, 30 per cent) than in thin patients whose BMI was <24 (41 of 288, 14 per cent; χ2 = 13·8, P < 0·001). This study failed to show any advantage to starting antibiotics pre-operatively, even in appendicitis.