The most obvious achievements of clinical science have been in the elucidation of symptoms and signs and the patterns of disordered function due to failure of the organs or to nutritional disturbances. The benefits of clinical research are both direct--through improved practice--and indirect--through improved teaching and contributions to biological science. It is suggested that the clinical scientist, experienced in both clinical and research work, has a potential not to be expected from collaboration between non-scientific clinicians and non-clinical scientists. Problems which particularly affect clinical research include: ethics; difficulty in being experimentally rigorous; the need to be opportunistic; dependence on transient workers; excessive conern with the end stages of irreversible disease; triviality; uncritical and premature imitation of research in practice. Clinical science is always threatened by a tendency for its problems to be regarded simply as applied problems of basic science. The roles of the clinical scientist should include: elucidating clinical phenomena, about most of which we remain very ignorant; collaboration with basic scientists on the one hand and with community scientists on the other; and clarifying the description and analysis of illnesses.