OBJECTIVES: To analyze the difference of medical expenditures between Angiotensin II receptor blockers (ARB) users and the users of other antihypertensive drugs for hypertension after ARB adopted for reimbursement by the Bureau of National Health Insurance in Taiwan since February 1998. METHODS: This retrospective study used the sampled cohort National Health Insurance claim database from 1997 through 2004, and selected the patients of the first taking antihypertensive drugs since February 1998, and divided the patients into ARB groups and non-ARB groups by medicine proportion of days covered (PDC) over three years. Various medical expenditures were calculated for comparison, including total expenditure, drug expenditure, non-drug expenditure, outpatient and inpatient expenditure and cardiovascular relative expenditure. Multiple linear regression and quantile regression was used to analyze the relationship between utilization of ARB and medical expenditures by controlling compliance, patient characteristics, and hospital accreditation levels. RESULTS: There were 7978 patients in our analysis, including 135 ARB users and 7874 non-ARB users through our exclusion criteria, including the principle of medicine proportion of days covered more than 70%. The results displayed that ARB users' three-year average total expenditure, drug expenditure and non-drug expenditure per patient were higher than those of non-ARB users. Though there was a negative impact of ARB on q0.75 non-drug expenditure, this coefficient was not statistical significance. The compliance indicator of ARB users was better than that of non-ARB users, excepting angiotensin converting enzyme inhibitors. CONCLUSIONS: ARB users' total medical expenditures were higher than those of other antihypertencive drugs users, but ARB users have better compliance. We suggest that ARB can be used for the patients whose intolerance of other antihypertensive drugs is not satisfactory or for the indication that ARB could provide better treatment, like hypertension with diabetes.