Although random assignment is generally the preferred methodology in impact evaluations, it raises numerous ethical concerns, some of which are addressed by securing participants' informed consent. However, there has been little investigation of how consent is obtained in social experiments and the amount of information that can be conveyed--and absorbed--prior to consent. This article reports on the implementation of the Employment Retention and Advancement (ERA) program, a large social experiment in the United Kingdom. Procedures to elicit informed consent were generally scrupulously followed, but even so many participants appeared to have only limited understanding of the experiment and their involvement in it. The reasons for this are articulated and recommendations are made to help ensure, to the extent possible, that potential random assignment study participants adequately understand their situation and their choices. (Contains 1 table and 1 figure.)