Patients can be harmed by treatment or by the decisions that are made about those treatments. Although dramatic examples of harmful effects of psychotherapy have been reported, the full scope of the problem remains unclear. The field currently lacks consensus about how to detect harm and what to do about it when it occurs. In this article, we define the ways in which treatment (or the inferences about treatment) can do harm and discuss factors that complicate efforts to detect harm. We also recommend methods to detect and understand harm when it occurs, drawing from and modifying many of the same strategies that are used to detect benefit. Specifically, we highlight the value of establishing independent systems for monitoring untoward events in clinical practice, reporting descriptive case studies and qualitative research, and making use of information from randomized clinical trials, including examining potential active ingredients, mechanisms, moderators, and a broad range of outcomes measured over time. We also highlight the value of promoting discussion in the field about standards for defining and identifying harm. (Contains 1 table.)