Research regulations around the world do not impose any limits on the risks to which consenting adults may be exposed. Nonetheless, most review committees regard some risks as too high, even for consenting adults. To justify this practice, commentators have appealed to a range of considerations which are external to informed consent and the risks themselves. Most prominently, some argue that exposing consenting adults to very high risks has the potential to undermine public trust in research. This justification assumes that it is not the magnitude of the risks themselves which raises concern, but the way in which the public might respond to them. This justification thus depends on the possibility that the public will find out about the risks and respond to them in the specified way. Like the other proposed external justifications, it thereby fails to offer a reason to think that exposing consenting adults to very high risks is problematic in itself. In the present paper, we describe and endorse a different justification. Rather than appealing to external factors, we argue that limits on risks for consenting adults trace to internal limits on informed consent, to limits on the things consent can and cannot make ethically permissible. In doing so, we aim to provide a firmer conceptual basis for the view that some research risks are unacceptably high, no matter how the research is conducted.