OBJECTIVES: To model the 1-year cost-utility of rabeprazole and esomeprazole "on demand" (prn) treatment for symptomatic gastro oesophageal reflux disease from the perspective of the UK National Health Service. METHODS: Data relating to treatment discontinuation due to inadequate heartburn control were extracted from two clinical trials; one comparing rabeprazole 10mg with placebo prn and the other comparing esomeprazole 20mg, 40mg and placebo prn. Survival data (proportion of patients continuing therapy) were fitted to Weibull functions, and adjusted for comparability according to placebo data. Data from the trials on drug intake, use of antacids as rescue medication and severity of heartburn symptoms were also used for the analysis. Health care resource utilization included annual frequency of general practitioner and gastroenterologist consultation and of upper GI endoscopy, annual number of drug prescriptions and pharmacy dispensing fees. These were priced according to the latest NHS costs. Health state utilities were derived from a study that assessed EQ-5D utilities in 1003 patients with GERD, and related utility scores to duration and severity of symptoms. A probabilistic model was employed that sampled from Weibull distributions for survival time, assigned Poisson distributions to annual frequency of events, Beta distributions to utilities and Dirichlet distribution to severity of heartburn. RESULTS: The mean total costs of therapy with rabeprazole 10mg, esomeprazole 20mg and 40mg were £93, £103, and £121, respectively. The associated utility scores were, respectively, 0.866, 0.861 and 0.860. CONCLUSIONS: For non-erosive reflux oesophagitis, treatment with rabeprazole 10mg prn is less expensive than with either 20mg or 40mg esomeprazole prn. All three alternatives are comparable in terms of their effectiveness.