[Purposes] When the Taguchi method is used to improve product reliability, the observations of lifetime (as response) may be censored while the Taguchi data analysis method is not applicable for such data. A new method is needed to address this issue. [Methods] A Weibull approximation approach is proposed to estimate the median residual life of a right-censored observation, and the sum of censored time and estimated median residual life is used to replace the censored observation. In this way, incomplete data are transformed into complete data so that the Taguchi data analysis method can be used. [Findings] Two simulation numerical examples are used to illustrate the accuracy and robustness of the proposed method and a real-world example is used to illustrate its usefulnesss. [Conclusions] Thus, the paper provides a simple and effective method for analyzing the data from reliability improvement experiments.