OBJECTIVES: To assess the impact on the health care expenditure and prevention of hospitalization of cardiovascular disease (CVD) for beneficiaries of Taiwan’s National Health Insurance (NHI) scheme after rosuvastatin (RSV) become available in NHI’s benefit package in May, 2005. METHODS: Using NHI’s claim data of a nationally representative random sample (consisting of 1 million of eligible beneficiaries) from 2004-2008, we classified patients >= 20 years of age who received other statins before and switched to RSV during May 2005 to April 2006 into switch group (N=1,040). Other patients who received statins therapy except RSV before May 2005 and throughout 2008 were classified into control group. Using propensity score (PS), we selected match group (N=1,039) from control group (N=18,432) based on age, gender, prior ambulatory care expenditure, and comorbidities related to the probability of switching. The date of first claim for prescribing RSV or other statins after May 2005 was the index date. Cumulative health expenditure after 1, 2, and 3 years from index date were compared between switch group and match group. Cox regression model controlling for age, sex, and comorbidities was applied to investigate the relative risk of hospitalization for CVD between two groups. RESULTS: Patients who switched to RSV are significantly consumed more ambulatory care resources after 1, 2, and 3 year of switching ($348, $680, and $1,050 more, respectively). Before PS matching, the relative risk of hospitalization CVD in switch group is significantly higher than control group(RR=1.38, 95% CI=1.12–1.70). After PS matching, the relative risk is still higher than match group (RR=1.18, 95% CI=0.881.58), but with less significance.  CONCLUSIONS:  RSV may be more cost-effective in lowering low-density lipoprotein cholesterol, but when taking the prevention of CVD hospitalization into account, the cost-effectiveness is diminishing. Putting new drugs into coverage should be considered more rigorously when cost is higher but benefit is not substantive.