People with mental illness are treated, in research, as a ‘class’ or category who are vulnerable, without always being clear why they should be treated as such, not why an individual, rather than the class, is vulnerable. The two main reasons given are lack of competence and power imbalance. Competence issues include incapacity and legislation, assessment and the impact of the illness in decisions. Power issues cover the role of mental health legislation, coercion, protectiveness and paternalism, stigma and discrimination and the ‘well’ and ‘ill’ self. These are examined within a framework of risk assessment, both to the patient and to others. Over-concern to protect the vulnerable can lead to gate-keepers preventing their inclusion in studies at various stages of the research. This denies such groups a voice in the research process. It is suggested that ethical scrutiny of research is not only to prevent harm to the vulnerable, but also to provide a framework to empower such people to take part in research. To do otherwise is to further stigmatize and marginalize them. One way forward is to look at the risk presented by different types of research, from clinical intervention trials to non-therapeutic, qualitative studies and to modify capacity and access requirements in the light of the risks presented.