In clinical trials, patient outcomes are often measured at scheduled time intervals. Treatment effects at a pre-specified time point are often tested. Patient dropout may introduce great uncertainty in the inference of treatment effect. In particular, the possible informative dropout makes it difficult to choose statistical models for reliable inferences. Dropout mechanisms are often hard to characterize by statistical models. Therefore, comparisons of the available statistical methods are difficult to make. In this presentation, we consider a set of commonly known empirical dropout models. Simulation studies are performed to evaluate and compare the reliability of these empirical models with various patient dropout mechanisms including informative dropouts. Based on these results, we will give suggestions regarding the applicability of these empirical models in a few therapeutic areas.