Mucosal dysplasia has been used as a marker for patients with chronic ulcerative colitis considered to be most at risk of developing cancer, and its identification is the basis for colonoscopic surveillance programs. To evaluate the reliability of this premise, colectomy specimens from two groups of patients who had undergone surgery for chronic ulcerative colitis (50 with cancer and 50 without) were retrieved. The groups were matched by age, sex, duration of disease, disease extent, and symptoms at the time of surgery. Using a standard technique of multiple random biopsies, we utilized the standard colonoscopic biopsy forceps to obtain four biopsies from mucosa that was not macroscopically suspicious for dysplasia or cancer in eight defined regions in each of the 100 colon specimens. This technique mimicked exactly the methods used in our clinical surveillance program. All 3,200 biopsies were evaluated blindly by one pathologist for presence and grade of dysplasia. Twenty-six percent of colons with an established cancer harbored no dysplasia in any biopsy from any region in the colon. While an overall association between the presence of cancer and high-grade dysplasia was detected (relative risk = 9.00; 95 percent CI of 2.73-29.67), the sensitivity and specificity of random colonic biopsies to detect concomitant carcinoma were 0.74 and 0.74, respectively. These findings prompt concern that reliance on random biopsies, obtained during colonoscopic surveillance, may be misplaced.