BACKGROUND Normalization of gastric secretion and cure of associated upper gastrointestinal lesions by resection of gastrinoma is possible in approximately 20% of patients with Zollinger-Ellison syndrome, leaving approximately 80% dependent on medical treatment with proton pump inhibitors for acid suppression. METHODS Lansoprazole was given for 3-48 months (median 28 months) to 26 Zollinger-Ellison syndrome patients with peptic ulcer manifestations in all and oesophagitis in 13. Starting with 60 mg/day. the dose was individualized to lower basal acid output to less than 5 mmol/h for those with intact stomachs and less than 1 mmol/h in those who had prior gastrectomy or with oesophagitis. The patients were studied every 3 months for 1 year and then every 6 months with gastric analysis (basal and maximal acid and pepsin output) and endoscopy with biopsy for enterochromaffin-like (ECL) cells. RESULTS Lansoprazole inhibited basal acid output by 95%, pepsin output by 65% and remained effective at the initial mean (66 +/- 4.3 mg/day) or smaller doses (56 +/- 12 mg/day) at 48 months. Mucosal lesions healed and symptoms (ulcer-type pain, diarrhoea, heartburn, weight loss) resolved rapidly, usually within a few weeks. Serum gastrin and ECL cell populations, which were elevated before treatment, remained statistically unchanged but one of the three multiple endocrine neoplasia I (MEN-I) patients developed a small carcinoid. Of the three patients with metastatic gastrinoma at diagnosis one has died and one has progressed, while the third has had stable liver metastases for 26 years. Ulcer-type relapses occurred in three of the five post-gastrectomy patients, one with fatal jejunal ulcer perforation despite adequate acid suppression. No biochemical or clinical adverse events due to lansoprazole were encountered. CONCLUSIONS Lansoprazole effectively inhibits acid and pepsin secretion in Zollinger-Ellison syndrome patients without any demonstrated side-effects. Despite strict acid control, post-gastrectomy Zollinger-Ellison syndrome patients were more liable to ulcer relapse, while oesophagitis was not a marker for therapeutic difficulty.