Missing data is a significant issue in clinical trials conducted in areas such as pain, depression, schizophrenia, and others in CNS. For the most part, it arises from patient withdrawal/dropout and could be associated with dropout as high as 60%. We will talk about the impact of this issue on estimating treatment differences and will use data coming from a schizophrenia trial. We will examine dropout patterns, evaluate determinants of propensity to drop out, and estimate treatment differences based on traditional approaches (LOCF and BOCF) in addition to more recent methods (MMRM, Pattern Mixture Models, Propensity-score-based models, and selection models).