Abstract
One of the central ethical requirements for medical research is that the risks to individual research participants should not be excessive. Despite widespread endorsement of this requirement, how it should be implemented remains controversial. This paper critically reviews the four existing approaches to evaluating the risks and benefits of medical research. It argues that the ‘net risks test’ is superior to ‘component analysis’, the ‘agreement principle’, and the ‘integrative approach’ and ends by highlighting several important questions that remain to be addressed.