• We compared two methods of estimating parietal bacterial contamination during abdominal operations. Duplicate swabs were taken from the subcutaneous tissues at the end of 817 operations; one was transported to the Department of Microbiology in Stuart's thioglycollate medium and the other immediately incubated in Robertson's cooked meat broth in the operating room suite and subsequently subcultured. The broth cultures revealed significantly more isolations of potentially pathogenic bacteria and more accurately predicted the likelihood of wound infection. In particular, when visceral cultures were positive, broth culture of wound swabs predicted a major wound infection rate of 0% when sterile, 4.8% when a single pathogenic species was cultured, and 10.1% when two or more were cultured. The corresponding figures for thioglycollate-transported swabs were 1.0%, 10.5%, and 12.9%. We conclude that broth cultures of parietal swabs allow accurate identification of patients at risk of infection from bacterial contamination of the wound during operation. (Arch Surg 1987;122:33-37)