When conducting clinical trials, every effort should be made to achieve complete capture of all data. In practice, however, some missing values can arise for various reasons in a clinical trial, especially for diagnostic devices. For handling missing data, there are different methods depending on the underlying missing mechanism. From regulatory perspective, the primary method of handling missing data should be justified based on the expected data mechanism. To ensure that the diagnostic study results are not driven by way of handling missing data, alternate methods for imputing missing data based on different assumptions are recommended to evaluate the robustness of the study results. In this presentation, the speaker will share the experience of handling missing data in in vitro diagnostic device. Different methods are used for handling missing data to assess the robustness of the device performance. The interpretation of corresponding analysis results will be discussed.