This study compared the sensitivity and specificity of CA 50 and CA 19-9 as serum tumour markers for pancreatic cancer. One hundred and seventy one subjects were evaluated: 50 healthy controls, 50 patients with pancreatic carcinoma and 71 patients with chronic pancreatitis. Eighty per cent of the pancreatic cancer patients had raised CA 19-9 serum levels and 82% had raised CA 50 serum levels. In the group of patients ith chronic pancreatitis, false positive tests occurred in 8.4% for CA 19-9 and 11.3% for CA 50. For both markers the serum level showed a severe elevation in the advanced stage of cancer disease. Despite the good sensitivity and specificity of CA 50 as a serum tumour marker for pancreatic carcinoma, no major advantage was found compared to CA 19-9. Moreover, if the CA 50 cut-off, level is raised to 85 U/ml to exclude overlap with chronic pancreatitis, sensitivity falls sharply to 46% compared with 72% for CA 19-9 when a 100 U/ml cut-off level is used.