OBJECTIVES: To determine proton pump inhibitor (PPI) utilization and their effect on treatment costs in patients with gastroesophageal reflux disease (GERD). METHODS: Data were extracted from National Health Insurance Health Care claims database in Taiwan. We identified continuously enrolled patients diagnosed with GERD and newly treated with a PPI between Jan 1, 2001 and Dec 31, 2001. Data were analyzed for six months following PPI initiation. Results were stratified by first PPI filled during the study period. GERD severity groupings were measured using a 4-step scale ranging from an endoscopic diagnosis of mild esophagitis to stricture. Compliance (as measured by a medication possession ratio), length of therapy (using total days supplied ) and daily average consumption (as measured as number of units consumed per day) were measured. Regression analysis was performed on GERD-related costs using treatment patterns, type of PPI drug, and compliance as independent variables of interest. RESULTS: Of 1600 remained subjects for study, there are 790 (49.4 %) omeprazole, 666(41.6%) lansoprazole, and 144 (9.0%) pantoprazole subjects. The possession ratio was not significantly different between drug subjects. The mean (median) length of therapy of omeprazole subjects was 76 (56) days, 55(42) days for lansoprazole subjects and 56(49) days for pantoprazole subjects. Among subjects pantoprazole users did not have a lower daily average consumption (1.00) versus both lansoprazole (1.01) and omeprazole subjects (1.03, p=0 .057). Compliance with PPI therapy had 45.1% higher GERD-related pharmacy costs (p < .001) and 40.4% higher GERD-related total costs (p < .01). Omeprazole subjects had 17.8% higher GERD-related pharmacy costs versus pantoprazole subjects (p < .001). Lansoproazole subjects had 12.7% higher GERD-related total costs versus pantoprazole subjects (p < .01), while Omeprazole subjects had 24.9% higher GERD-related total costs versus pantoprazole subjects (p < .001). CONCLUSIONS: Compliance was not significantly different between the PPIs, nor did increased compliance decrease GERD-related costs. Pantoprazole subjects had lower GERD-related costs, compared to lansoprazole and omeprazole subjects.