The recommended approach to the increased risk of colorectal carcinoma in ulcerative colitis has been colonoscopic surveillance rather than prophylactic colectomy. This strategy is based on the assumption that dysplastic lesions can be detected before invasive cancer has developed. We have analysed published reports on dysplasia surveillance to find out whether this assumption is valid. Ten prospective studies (1225 patients) satisfied our criteria. Of 40 patients with dysplasia-associated mass or lesion (DALM) detected, 17 (43%) already had cancer at immediate colectomy. The risks of cancer at immediate colectomy were 42% (10 of 24 patients) for high-grade and 19% (3 of 16) for low-grade dysplasia. Of 47 patients found to have high-grade dysplasia after the initial colonoscopy, 15 (32%) had cancer. 16-29% of patients with untreated low-grade dysplasia progressed to DALM, high-grade dysplasia, or cancer. Of patients with indefinite results, 28% progressed to high-grade dysplasia and 9% to cancer, so continued surveillance is essential. The risk of progression to dysplasia was only 2.4% for patients whose initial result was negative, so surveillance could perhaps be less frequent for these patients. Immediate colectomy is essential for all patients diagnosed with high-grade or low-grade dysplasia. A diagnosis of dysplasia does not preclude the presence of invasive cancer. We believe that patients should be informed about the limitations of colonoscopic surveillance so that they can take part rationally in decision-making about their management.