It may be useful to draw an analogy between the proposed screening programme for colorectal cancer and the cervical cancer screening programme. Both tumours show a spectrum of histological abnormalities consistent with a premalignant phase. The natural history of these premalignant lesions is poorly understood and although some will progress, if untreated, to invasive disease, most will not. Light microscopy cannot confidently distinguish which cases will progress and which will regress, and clinicians are therefore obliged to treat all. This will result in the destruction of many lesions of uncertain malignant potential. The secondary prevention of cervical cancer, although therapeutically efficacious, is inefficient. A lack of understanding of the natural history of intraepithelial neoplasia has frustrated attempts to develop rational referral criteria, and it is only now that the appropriate trials are being undertaken. The development of outpatient investigative and therapeutic procedures has resulted in many more women being referred for investigation and treatment, with predictable pressure on other services offered by gynaecologists, but no demonstrable saving of life. Similar uncertainties surround a screening programme for colorectal cancer. The principal concerns are not about the efficacy of polypectomy in interrupting the polyp cancer sequence, although uncertainties about the frequency with which cancer arises de novo do require that the effectiveness of this intervention is formally tested. Our major concerns are with compliance, and the management of the individual who tests positive--that is, who is found to have a distal polyp. Technological advances and operator enthusiasm may, as has happened with the cervical screening programme, lead to a relaxation in the indications for further investigation and treatment. Such a development would affect resources substantially if a population screening programme were in place. Nevertheless, there are grounds for believing that a screening programme for colorectal cancer, using sigmoidoscopy, might be successful in certain age groups if compliance was satisfactory. The scale of benefits may be comparable with those achieved by the breast screening programme. Our limited cost analysis, which relates to only to specific items of clinical activity, suggest that the mean cost for each case of cancer prevented will be about 8000 pounds sterling. These conclusions suggest that screening by flexible sigmoidoscopy merits serious consideration. It is also imperative, however, that consideration should be given to resolving some of the uncertainties about the clinical management and surveillance of those found to have distal polyps.