Detection of cancer is the definitive endpoint in the conduct of chemoprevention trials. There are, however, several reasons why cancer as the endpoint may not be feasible or ethical: 1) the usage of patients with easily followable preneoplasias may preclude the development of cancer, and 2) the time to a cancer event may be long or the incidence uncommon, even in individuals at high risk. Two major types of biomarker intermediate endpoints should be considered: 1) those that identify individuals at high risk and 2) those that serve as a surrogate for cancer. Various epidemiologic features, including family history, have been used to estimate relative risk. This approach, however, only slightly decreases the size of populations needed for chemoprevention trials and only little addresses the question of individual risk. Advances in understanding the genetic basis for cancer will lead to the development of probes that will help assess risk for many cancers. Innumerable biomarker intermediate endpoints can be identified as associated with cancer formation, including genetic, epigenetic, and histologic features. The challenge is not in identifying potential biomarker intermediate endpoints but in showing that they are relevant. Carcinogenesis has been shown to be carcinogen, inhibitor, dose, tissue, and species specific; it is likely that relevant biomarker intermediate endpoints will need to be identified, studied, and verified in human models. The upper aerodigestive system should be a rich source for biomarker intermediate endpoint studies, as tissue is readily available, the carcinogenic process can be monitored, and there are currently available reasonable compounds to use in biomarker intermediate endpoint modulation and chemoprevention trials.(ABSTRACT TRUNCATED AT 250 WORDS)