OBJECTIVES: To review the issues in calculating confidence intervals (CIs) of the incremental cost-effectiveness ratio (ICER) with conventional methods and to propose new methods defining more appropriate ranges to evaluate the uncertainty of cost-effectiveness analysis. Conventional methods to calculate CIs first obtain the cost-effectiveness plane through Monte Carlo simulation. Next, 95% of the distribution is surrounded with two lines that pass through the origin, and the slopes of these two lines will be assumed to be the upper and lower CIs. In some situations, the methods yield CIs that are separated as -∞to upper limit and lower limit to ∞. On the contrary, CIs are not defined if any two lines surround 95% of the distribution. METHODS: We propose two modifications on methods for constructing ICER CIs. One is to surround 95% of the distribution with two segments from the origin (method 1). Another is to report just the upper or lower limit defined in the first quadrant if the CIs are not appropriately defined (method 2). We evaluated the performance of the existing and proposed methods (method 1, method 2 and a combination of both) using simulations that calculate ICER CIs on three situations. RESULTS: Our simulations revealed that the results from the proposed methods are similar to those from the conventional ones under the situation that the latter can yield the fair CIs (situation 1). However, under the situations that conventional methods cannot appropriately define the CIs (situation 2; CIs separated, situation 3; CIs cannot defined), the combination of the two proposed methods produces CIs as single range. CONCLUSIONS: We reviewed the issues related to using the conventional methods to calculate ICER CIs, and proposed new methods to address these. Proposed methods yield CIs that can be more easily interpreted. However, some issues remain that require further discussion.