One hundred forty-nine patients aged 40 years or more with symptoms suggestive of colonic disease but without evidence of gastrointestinal bleeding (absence of hematochezia, normal serum levels of hemoglobin, and at least one test negative for fecal occult blood) were randomized to undergo either initial colonoscopy or initial flexible sigmoidoscopy plus air-contrast barium enema. Patients with incomplete initial colonoscopy and certain patients with polyps seen on flexible sigmoidoscopy plus barium enema underwent the alternative procedure (barium enema or colonoscopy). The main results were as follows: First, the overall prevalence of cancer in the study was very low (0.67%). Second, initial flexible sigmoidoscopy plus barium enema detected more patients with diverticulosis than did initial colonoscopy (46% versus 31%; p = .01). Initial colonoscopy detected more persons with adenomas (p = .06) than did initial flexible sigmoidoscopy plus barium enema. Patients undergoing initial flexible sigmoidoscopy plus barium enema require the alternative procedure (24%) than were patients undergoing initial colonoscopy (6%; p = .002). Third, sensitivity analyses suggested that for most areas in the United States, initial colonoscopy would be more cost-effective for the outcomes of detection of adenomas and detection of large adenomas, although very few patients in the study had large adenomas. We conclude that the prevalence of colorectal cancer in persons with colonic symptoms but no evidence of bleeding is low and is comparable with the prevalence in an asymptomatic population. Cost-effective selection of imaging strategies in this population can be based on demographic factors such as age and sex, which are better predictors of the presence of adenomas than are symptoms.