The results of a prospective study designed to investigate patients suspected of having pancreatic cancer are reported. One hundred and two of 238 patients investigated had pancreatic cancer. Ultrasonography, endoscopic retrograde cholangiopancreatography, and cytology were the most reliable tests for the diagnosis of resectable cancer. Computed tomography had a higher sensitivity for unresectable cancer. Factors responsible for delay in diagnosis are discussed. Cancer of the body and tail of the pancreas cannot be diagnosed early by investigating a symptomatic population. Factors influencing mortality in patients who survive over three years following pancreatic resection are discussed.