The use of change-from-baseline endpoints is pervasive in clinical drug development, most noticeably in confirmatory trials. This leads to inefficient use of longitudinal information collected during the trial (with just two time points being used), as well as potential difficulties with dealing with missing data. Key motivations for the change-from-baseline endpoint are that it does not require knowledge about the time-response profile of the underlying endpoint and has a straightforward clinical interpretation. This talk will discuss the use of parametric time-response models for longitudinal data, in the context of mixed-effects models. The appropriateness, benefits and challenges of the approach will be illustrated with real drug development applications.