Compared particularly with Scandinavian countries, the effects of cervical screening in the United Kingdom have been disappointing, both in terms of women screened and in relation to incidence and mortality. The underlying problem is the under-representation of women most at risk of cervical cancer in the screened population. In this paper we examine two possible hypotheses for such under-representation: one concerns the behavioural aspects of women's failure to attend and the second the effect of the organisation of current screening services on their attendance. We outline the features that an effective service would need to incorporate and propose principles for the development of a screening system designed to maximise the potential of the cervical smear test.