BACKGROUND/AIMS Cancer surveillance in patients with ulcerative colitis is of unproven benefit. This study assesses the efficacy and analyzes factors limiting the success of a surveillance program during a 21-year period in 332 patients with ulcerative colitis to the hepatic flexure and disease duration exceeding 10 years. METHODS Clinical assessment and sigmoidoscopy with biopsy was undertaken yearly. Colonoscopy and biopsy every 10 cm throughout the colon was performed every 2 years or more often if dysplasia was found. Only biopsy specimens reported as showing dysplasia were reviewed. RESULTS Surveillance contributed to detection of 11 symptomless carcinomas (8 Dukes A, 1 Dukes B, and 2 Dukes C), but 6 symptomatic tumors (4 Dukes C and 2 disseminated) presented 10-43 months after a negative colonoscopy. Dysplasia without carcinoma was confirmed in 12 symptomless patients who underwent colectomy. The 5-year predictive value of low-grade dysplasia for either cancer or high-grade dysplasia was 54% using current criteria. CONCLUSIONS Surveillance identified some patients at a curable stage of cancer or with dysplasia. Limiting factors were failure to include patients with presumed distal colitis, biennial colonoscopy, the number of biopsy specimens at each colonoscopy, and variation in histological identification and grading of dysplasia.