One of the arguments supporting the concept that benign adenomatous polyps of the colon degenerate into cancer is the observation of malignancy in polypoid growths known to have been present for years. Although a latent phase-the duration of time between the initial development of a malignancy and the subsequent occurrence of a clinical problem-must exist, this argument implies that this time span is not many years. If both the prevalence of malignant polyps and the incidence of consequent symptomatic colon cancer were known, the average latent phase could be calculated. In order to estimate this prevalence, I used autopsy data; the estimate was validated using independent data from three colonoscopic screening studies. The annual incidence of all colon cancer in the United States is approximately 150 000 cases. I estimate that 725 000 people in the United States harbor at least one malignant polyp. Even if all 150 000 cases of colon cancer were associated with symptoms and began as malignant polyps, the average latent phase is 4.8 years. Because some colon cancers are removed while the patient is still asymptomatic (discovered on screening examination) and because at least some colon cancers arise de novo from the mucosa, the average latent phase must be even longer. These estimates suggest that it cannot be assumed that the histologic finding of cancer in a polyp that has been observed for many years represents ''malignant degeneration'' of a previously benign neoplasm; such a malignancy may have had that histologic characteristic from the start. Further, histologically ominous lesions (malignant polyps) may often have prolonged benign clinical courses.
