The records of 300 consecutive patients who underwent cholecystectomy at a large military teaching hospital from January 1981 through August 1982 were reviewed; 270 patients (90 per cent) had intraoperative cholangiograms. Four clinical indications were helpful in predicting the likelihood of a positive intraoperative cholangiogram: 1, jaundice; 2, pancreatitis; 3, dilated common bile duct, and 4, palpable stones. The 63 patients with at least one of these indications had a 35 per cent incidence of true positive cholangiograms, while the 207 patients without such indications had a 0.5 per cent incidence of true positive studies (p less than 0.01). The incidence of false-positive studies was 0.7 per cent in our series and 3.1 per cent in our review of the literature of 2,580 cholangiograms. We conclude that cholangiography in the absence of clinical indications has a low yield. If cholangiography had been used selectively during the time period of 20 months of our retrospective study, more than 25,000 dollars could have been saved without missing significant pathologic findings in the common bile duct. We conclude that the thoughtful surgeon should perform cholangiography on a selective rather than routine basis.