The diagnostic accuracy of ultrasound (US), computed tomography (CT), endoscopic retrograde cholangiopancreatography (ERCP), and tumour markers CEA, CA 50 and CA 242 in pancreatic cancer (n = 26) was studied in 113 patients with jaundice, in 20 patients with unjaundiced cholestasis, and in 60 patients with the suspicion of chronic pancreatitis or a pancreatic tumour. The sensitivities of US, CT and ERCP were 61.9%, 95.2% and 82.3%, the specificities 93.9%, 92.9% and 94.1%, and the efficiencies 91.6%, 96.6% and 92.1%, respectively. The sensitivities of CEA, CA 50 and CA 242 were 92.3%, 96.1% and 61.5%, the specificities 59.2%, 58.0% and 95.2%, and the efficiencies 63.7%, 63.2% and 90.6% respectively. The combined use of the imaging methods and tumour markers was also analysed. When either the imaging method or the serum marker test was required to be positive, the sensitivities of the combinations were clearly better than those of US and CA 242 alone, but only slightly better than those of CT, ERCP or the tumour markers CEA and CA 50 alone. When both the imaging test and the marker test were required to be positive, the specificities of the combinations were clearly better than those of CEA and CA 50 alone, but they did not exceed the specificity of the imaging methods or CA 242 alone. We conclude that CT, ERCP and CEA and CA 50 are highly sensitive in the diagnosis of pancreatic cancer in symptomatic patients, while the sensitivity of US and CA 242 is lower. The specificity of the imaging methods and CA 242 is high, but that of CEA and CA 50 is low. Imaging methods and serum tumour markers could be more used in clinical practice in a complementary manner. In patients with jaundice and/or cholestasis or with a suspicion of pancreatic tumour or chronic pancreatitis, the combined use may yield higher sensitivity than US alone and higher specificity than CEA or CA 50 alone.