Gastric aspirates were obtained from 12 healthy volunteers, 49 patients with duodenal ulcer, 14 with gastric ulcer and 35 with gastric carcinoma. The mean total viable bacterial counts in these groups were as follows: volunteers 0, duodenal ulcer 3.8 X 10(1), gastric ulcer 6.95 X 10(4), carcinoma 1.9 X 10(7) organisms/ml. The incidence of wound sepsis in patients without antibiotic cover was; duodenal ulcer 17 per cent, gastric ulcer 38 per cent, carcinoma 56 per cent. Regardless of the underlying pathology, patients with counts greater than 5 X 10(6) organisms/ml in the gastric aspirate had a 93 per cent incidence of wound sepsis, compared with 16 per cent in patients with counts of less than 5 X 10(6) organisms/ml (P less than 0.001). In the group with high counts all except one of the wound infections were caused by organisms present in the stomach at the time of operation. There was a good correlation in the bacteriology of apirates obtained during preoperative endoscopy compared with operative nasogastric samples (n = 31) both for viable counts (r = 0.93) and for the counts of individual organisms. Therefore, preoperative endoscopy can be used to identify patients who are at risk of developing wound sepsis after gastric surgery.