CA 19.9 serum levels were prospectively determined in 573 patients admitted to hospital for suspicion of pancreatic cancer. The final diagnosis was 77 patients with no malignancy, 389 patients with pancreatic cancer, 37 neuroendocrine pancreatic cancer, 28 cholangiocarcinomas, 4 gallbladder cancer, 27 ampullary carcinomas, and 11 periampullary carcinomas. CA 19.9 was determined using a commercial assay from Roche Diagnostics, and 37 U/ml was considered as the upper limit of normality. Abnormal CA 19.9 serum levels were found in 27%, 81.5%, 85.7%, 59.3%, 63.6%, and 18.9% of patients with benign diseases, pancreatic cancer, cholangiocarcinomas, and ampullary, periampullary, or neuroendocrine tumors. Significantly higher concentrations of CA 19.9 were found in patients with than in those without malignancy or with neuroendocrine tumors. CA 19.9 serum levels were higher in pancreatic cancer or cholangiocarcinoma than in other malignancies (p < 0.0001). CA 19.9 serum levels were also correlated with tumor stage, treatment (significantly lower concentrations in resectable tumors), and tumor location (the highest in those located in the body, the lowest in those in the tail or uncinate) and site of metastases (highest in liver metastases). A trend to higher CA 19.9 serum concentrations was found in patients with jaundice, but only with statistical significance in the early stages. Using 50 or 100 U/ml in patients with jaundice, CA 19.9 was useful as an aid in the diagnosis of pancreatic cancer (sensitivity 77.9%, specificity 95.9%) as well as tumor resectability in pancreatic cancer with different cutoffs according to tumor location and bilirubin serum levels with specificities ranging from 90% to 100%. CA 19.9 is the tumor marker of choice in pancreatic adenocarcinomas, with a clear relationship with tumor location, stage, and resectability.