Although staging of pancreatic cancer is critical to the planning of therapy, many patients come to laparotomy with liver or peritoneal metastases not suspected or detected during conventional preoperative testing. We performed laparoscopic examinations as part of the staging evaluation of 40 patients with proved pancreatic cancer who were candidates for curative resection or intraoperative radiotherapy. In each patient, laparoscopy was the last test before laparotomy and was carried out only if all other test results were negative for metastases. In 14 of 40 patients, single, small (1 to 2 mm) metastatic nodules were detected and verified by biopsy in the liver (6 patients), on the parietal peritoneum (7 patients), and in the omentum (1 patient). It is likely that several of these lesions would have been overlooked at routine exploration. Because of the positive findings, therapy was altered in all 14 patients. None received intraoperative radiotherapy, nine who did not have jaundice were discharged without operation, three had biliary or gastric bypass only, and two were treated by percutaneous biliary stenting. Of the 26 patients who had negative findings on laparoscopic examination, the absence of gross metastasis was confirmed at laparotomy in 23. Three false-negative findings were caused by incomplete examination of the liver in two patients and a central liver lesion in one patient. Therefore, 43 percent of all patients (17 of 40) had demonstrable but unsuspected nonlocal metastases, and laparoscopic examination detected 82 percent (14 of 17) of these. The overall accuracy of laparoscopy was 93 percent and that of examination with negative findings, 88 percent. In eliminating useless laparotomy and redirecting treatment plans, laparoscopy contributes significantly both to the proper management of patients with pancreatic cancer and to increased efficiency of resource utilization.