In many large bowel chemoprevention trials adenomas have a double duty: they are used to identify subjects at risk for large bowel neoplasia, and also serve as endpoints. Many features of adenomas make them suitable for these tasks. Patients with adenomas are fairly numerous and easy to identify; further, the 'adenoma-carcinoma' sequence suggests that adenomas are logical endpoints. The high recurrence risk among adenoma patients means that a relatively modest number of subjects will suffice for adequate statistical power. The are some limitations to the use of adenomas, however. There is clearly heterogeneity of risk for subsequent cancer. Patients with only small adenomas may have rates of colorectal cancer that are not much greater than those of the general population. Certainly subjects with large adenomas, and those with villous or highly dysplastic adenomas have a higher risk. Often, one would chose the high-risk patients for preventive interventions. Such a strategy makes sense from a risk-benefit point of view. However, from a population perspective, such a strategy may well have only a minor impact on the overall colorectal cancer burden. For more complete population-based prevention, efforts will have to be directed to the numerous individuals who are each at small risk, but who collectively account for most colorectal cancer. For this preventive approach, patients with any adenoma would certainly be part of the target population, and so are sensible subjects in chemoprevention trials. There are similar complexities in consideration of the use of adenomas as endpoints of chemoprevention trials. The adenomas that occur in prevention trials are generally small, and may not be associated with a greatly increased cancer risk. The issue for chemoprevention trials however, is not whether the endpoints are truly intermediate in the causal chain-but whether the intervention under study alters the adenoma recurrence risk to the same extent as it does for colorectal cancer risk. This is a difficult matter to verify, but the limited data available are encouraging. The epidemiology of colorectal adenomas (largely small adenomas) is similar in many regards to that for colorectal cancer itself. Thus to the extent that data are available, one can tentatively conclude that external influences affect adenomas and colorectal cancer similarly. To date, more than ten adenoma prevention trials have reported results. The data have been fairly consistent. Vitamin C (with or without vitamin E) has provided at most a modest protective benefit, except in one small trial in which it was combined with vitamin E and preformed vitamin A. beta-Carotene seems to be without any effect, and interventions to increase fiber and decrease fat intake have not indicated substantial effects. On the other hand, trials among familial polyposis patients have provided evidence for an impact of nonsteroidal anti-inflammatory drugs. Studies in progress have the potential to clarify greatly the preventive potential of the currently promising-but yet unproven-chemopreventive regimens.