In this paper we argue that the standard focus on problems of informed consent in debates about the ethics of human experimentation is inadequate because it fails to capture a more fundamental way in which such experiments may be wrong. Taking clinical trials as our case in point, we suggest that it is the moral offence of using people as mere means which better characterizes what is wrong with violations of personal autonomy in certain kinds of clinical trials. This account also helps bring out another important way in which the autonomy of the participants in clinical trials may be violated, even in cases where they have given informed consent to their involvement. Where relevant information about the trial is framed in such a way as to induce a patient's participation by appeal to their nonrational preferences, this is also a violation of their autonomy, and one which is distinct from a failure of informed consent. The underlying wrongness of both kinds of violations, we argue, is plausibly captured by the moral offence of using people as mere means.