OBJECTIVE: To determine whether patients with colorectal cancer are less likely than unaffected controls to have had one or more endoscopic procedures (flexible sigmoidoscopy, colonoscopy, or polypectomy) before being diagnosed with cancer. DESIGN: Case-control study. SETTING: Hospitals of the Department of Veterans Affairs. PATIENTS: 8722 and 7629 case-patients with colon and rectal cancer, respectively, and age-, sex-, race-matched controls who were discharged at the same time as the corresponding case-patients. MEASUREMENTS: Number and type of endoscopic procedures of the large bowel done from 1981 until the development of colorectal cancer in each case-patient. The influence of endoscopic procedures on the development of colorectal cancer was tested by conditional multiple logistic regression analysis. RESULTS: Compared with controls, patients with colorectal cancer were less likely to have had an endoscopic procedure of the large bowel before being diagnosed with cancer (odds ratio for colon cancer, 0.51 [95% CI, 0.44 to 0.58]; odds ratio for rectal cancer, 0.55 [CI, 0.47 to 0.64]). In patients who had flexible sigmoidoscopy, colonoscopy, and polypectomy, the odds ratios were even smaller. When analyzed by separate 1-year intervals, patients with cancer had significantly fewer procedures during periods of up to 6 years before their onset of their cancer. Similarly, fewer inpatient and outpatient procedures were done in patients than in controls. CONCLUSIONS: Endoscopic procedures of the large bowel reduce the risk for developing colon and rectal cancer by 50%, their protective influence lasting 6 years.