The management of pancreatic pain is a controversial subject and the treatment recommended varies from one extreme to the other. Some authorities advise simply waiting for chronic pancreatitis to 'burn out', while others practise removal of the entire gland. In this paper we present 141 patients who underwent surgery for chronic pancreatitis at the Mayo Clinic. The main indication for operation was pancreatic pain and the choice of operation was based on anatomical abnormalities in the gland. The long-term results of the policy are reviewed (mean follow-up 8.5 years). Length of history, aetiology of disease, pancreatic dysfunction and pathology, time after operation and continued alcohol abuse were computer analysed for a statistically significant influence on pain relief, ability to work, pancreatic function and survival. There was one operative death (mortality rate 0.7 per cent). Continued drinking was not shown to affect postoperative pain relief but 10-year survival was significantly less in alcoholics than in those with non-alcoholic pancreatitis (P less than 0.02). Dilated ducts and duct calculi were associated with good results for pain relief although this association did not achieve statistical significance. Parenchymal calcification and time after operation did not influence the results of surgery. When the operation failed to relieve pain, spontaneous remission occurred in a few cases only. Seventy-seven per cent of patients had lasting relief of pain and operations selected on the basis of gross pathology were equally effective in relieving pain. Longitudinal pancreaticojejunostomy in those with dilated ducts and a Whipple operation for disease of the pancreatic head gave good results.(ABSTRACT TRUNCATED AT 250 WORDS)