Abstract
National Health Insurance (NHI) has been implemented in Taiwan since 1995. This study's prime interest is to explore the Bureau's treatment on extreme values in the sampling audit and reimbursement process and see if the current method is reasonable or not. This research concludes: (1) there are significant relationships between the treatment of extreme values and the weighted sample-deduction ratios. (2) There are sufficient evidences to demonstrate that the current treatment of extreme values is misleading in the sampling audit and reimbursement process of BNHI. (3) The managers of NHI should identify extreme values before making reimbursement decisions.