The clinical value of the Lundh test in the diagnosis of pancreatic disorders was reviewed retrospectively for 139 patients with proven pancreatic disease (15 acute and 72 chronic pancreatitis, 51 carcinoma of the pancreas and one partial pancreatectomy). There was a good separation between patients with pancreatic disease and those with other gastrointestinal disorders who presented with similar symptoms. The test was most helpful in patients with chronic pancreatitis (diagnostic rate 89%, mean tryptic activity (MTA) < 7 U), more particularly in those presenting with jaundice or steatorrhoea (100%) than pain (86%), but less successful (diagnostic rate 68%) in carcinoma of the pancreas (steatorrhoea 100%; jaundice 64%; pain 55%). Complete absence of bile from duodenal juice in jaundiced patients usually indicated pancreatic or hepatobiliary carcinoma. An abnormal but not diagnostic MTA (7 to < 10 U) was seen in 7% of patients with chronic pancreatitis, 13% with carcinoma of the pancreas and 11% with non‐pancreatic disorders and suggested the need for further pancreatic investigation. Cytological examination of duodenal juice increased the usefulness of the test. Although malignant cells were rarely seen, characteristic (degenerate) cells were found in 22 of 44 (50%) patients with, but only five of 85 (6%) without pancreatic disease, even when the MTA was normal. The Lundh test can be conveniently combined with a jejunal biopsy at the same intubation for the investigation of diarrhoea or steatorrhoea.