In a county hospital, 217 patients were operated on for suspected cancer of the pancreas (186 patients) and periampullary structures (31 patients). The surgeons' philosophy was to avoid biopsy of the pancreas whenever possible, they often performed radical resection based on clinical judgment alone (8% of the resections were chronic pancreatitis). Follow-up of the 65 patients who were presumed to have carcinoma without histologic proof showed the surgeon's clinical diagnosis to be incorrect in over 7% of the cases (maximum error - 18%). Among patients with carcinoma of the pancreas, one third of the pancreatic biopsies missed the lesion (surgeon's sampling error). Frozen section examination of pancreatic biopsies in 63 patients showed false positive and negative rates of 3% and 4%, respectively (pathologist's sampling or interpretation errors). Two of 31 incisional biopsy of the pancreas resulted in pancreatic fistula; none of 24 needle biopsies did. Otherwise, biopsy of the pancreas did not influence mortality or morbidity after either resection or bypass procedures.