One approach often proposed to ensure that a treatment effect is both clinically meaningful and statistically significant is known as a "responder" analysis. In this approach, all subjects are classified as either responders or non-responders based on the value of a continuous outcome variable measured during the trial (or on a change from baseline in that variable). While the responder analysis approach has some appeal, it has disadvantages, such as the often arbitrary nature of the definition of a response and a loss of statistical power that can be considerable. In addition, the label “responder” implies that individual changes in the response variable are necessarily causal effects of the treatment. In this presentation we will give an overview of responder analysis commonly seen in different therapeutic areas, and then we will discuss the strengths and weaknesses of this approach.