In a randomized controlled trial (RCT), treatments are assigned randomly and treatments are withheld from participants. Is it ethically permissible to conduct an RCT in a social setting? This paper addresses two conditions for justifying RCTs: that there should be a state of equipoise and that the trial should be scientifically promising. Illustrated with a discussion of the RCT evaluation of the Wizards of Words reading programme, this paper argues that, first, the two conditions can give rise to genuine moral conflicts, and second, efforts can be made to ensure RCTs in social settings are scientifically promising. The argument of this paper therefore is a departure from the current debate on RCTs, where it is assumed these two justifying conditions should not come into conflict, either because research ethics is derived from the professional's duty of care, or because there is a strong distinction between the ethics of research and the duty of care. This paper also addresses critics who argue that in social settings RCTs cannot be scientifically promising and for that reason they are ethically impermissible.