The aims of this study were to (1) evaluate the diagnostic utility of a new tumor marker, TPS, with respect to TPA and CA 19-9 in patients with pancreatic cancer; (2) ascertain the reliability of the markers in predicting survival, and (3) evaluate the effect of liver dysfunction on the results. CA 19-9, TPA and TPS were measured in the serum of 19 control subjects, 42 patients with pancreatic cancer, 29 with chronic pancreatitis, and 52 with extrapancreatic diseases. CA 19-9 was confirmed to be the best serological indicator of pancreatic cancer, while TPA and TPS lacked both sensitivity and specificity. Pancreatic cancer patients with liver metastases had higher mean C A19-9 and TPA, but not TPS, values than pancreatic cancer patients without metastases. A shorter survival time was associated with the presence of liver metastases and with higher serum tumor marker levels. CA 19-9, TPA and TPS were found to be correlated with liver function test results (ALT, ALP and bilirubin). In conclusion: (1) TPS adds no significant information to that obtained using CA 19-9 in the diagnosis of pancreatic cancer; (2) CA 19-9 and TPA, but not TPS, are influenced by the presence of liver metastases; (3) the main factor to influence survival is advanced disease, which is in turn associated with higher tumor marker levels, and (4) liver dysfunction can influence not only CA 19-9 and TPA, as already described, but also TPS.