The controversial association of pancreatitis and pancreas divisum was studied in 100 patients (77 women, 23 men, median age 35) with episodic acute pancreatitis (49%) or "pancreatic pain" (51%). Seventy-one had classic pancreas divisum (type 1); 23 had only a dorsal duct with an absent Wirsung's duct (type 2); and 6 had a filamentous connection between the two duct systems (type 3). Accessory papilla sphincteroplasty was performed in 88 patients, with a mean follow-up of 53 months. The orifice was stenotic at the mucosal level in 66 patients. Seventy percent of patients have shown improvement: 85% if the accessory papilla was stenotic, compared with 27% if it was not (p less than 0.0001); and 82% with discrete attacks, compared with 56% with chronic pain (p = 0.002). Judged against intraoperative calibration of accessory papilla orifice diameter, ultrasonography with secretin stimulation was 78% sensitive for accessory papilla stenosis, with 3% false-positive results. Ultrasonography with secretin stimulation was the best predictor of surgical success: positive = 92% success (attacks or pain) versus negative = 40% success (64% with attacks; 21% with pain). There have been seven restenoses with six reoperations. We conclude that (1) pancreas divisum is but one variety of pancreatic anatomy characterized by a dominant dorsal duct and dependence on secretion through the accessory papilla; (2) accessory papilla stenosis appears to be a necessary cofactor to produce a morbid state, whether episodic pancreatitis or pancreatic pain; (3) presentation with pancreatitis and a positive result on the ultrasound-secretin test are the best predictors of successful accessory papilla sphincteroplasty.