OBJECTIVES: To assess the cost-effectiveness of two commonly used triple therapies for H. pylori eradication in a public hospital in Hong Kong. METHODS: A prospective observational design was adopted. Two cohorts of patients attending the GI Specialty Clinic at the Prince of Wales Hospital with endoscopically confirmed non-ulcer dyspepsia and H. pylori positive confirmed by CLO Test were recruited. Informed consent from patients and ethic approval from hospital were obtained. The patients were consecutively recruited to receive esomeprazole + amoxicillin + clarithromycin (EAC) or rabeprazole + amoxicillin + clarithromycin (RAC) therapies. One-week treatment was given and urea breath test was performed after 4 weeks to confirm if eradication was successful. Initial treatment failures were given quadruple rescue therapy (metronidazole + esomeprazole + danol + tetracycline). All health care resources including clinic visits, laboratory procedures, medications, and professional services were recorded. RESULTS: One-hundred and three and 96 patients were recruited to the EAC and RAC groups respectively. The demographic characteristics were not significantly different except the RAC group had a lower rate of drinking and smoking habits. The success rates for eradication for the EAC and RAC groups were 88.5% and 85.4% respectively (p=NS). The cost/successful eradication for EAC group and RAC group was USD846 and USD843.7 respectively. The cost/successful eradication after quadruple therapy was USD871 and USD868.3 for EAC and RAC groups respectively. The overall cost/patient for H. pylori eradication was USD849 and USD 847.3 for EAC and RAC groups respectively. No significant adverse effects were experienced by patients in either group and hence cost incurred in this area was minimal. CONCLUSION: Esomeprazole and rabeprazole triple therapies appear to have similar effectiveness in H. pylori eradication in patients with non-ulcer dyspepsia. Rabeprazole triple therapy may have cost savings when the whole population is considered.