The purpose of this paper is to clarify the short-term and long-term objectives of screening for various cancers, and to indicate the kinds of data that are needed to determine whether or not the objectives are met. Cancers at various sites differ with respect to their innate suitability for screening. Criteria that enhance screening suitability include the potential for serious complications and a high rate of mortality (applicable to most cancers), a prolonged preclinical phase, and an existing therapy that is simpler and more effective in reducing the mortality rate when applied to preclinical disease than to clinically evident cancer. Tests and procedures suitable for screening are simple to perform, inexpensive, acceptable to patients and physicians, safe, relatively painless, and accurate, as measured by the test's sensitivity and specificity. The actual yield of previously undiagnosed cancer arising from a screening program will depend heavily on prevalence of disease in the screened population, specificity of the screening test, and successful follow-up of screen-positive patients with diagnosis and treatment. These issues are discussed in the context of four cancers and their respective screening modalities: cervical cancer and cytologic studies, breast cancer and mammography, colon cancer and fecal occult blood tests, and lung cancer and sputum cytologic studies. The quality of data on which screening decisions have been made for each of these cancers and tests varies. The cancers vary in terms of their relevant biologic characteristics and treatment effectiveness. Similarly, each screening procedure has its own particular advantages and disadvantages. Current American Cancer Society Guidelines for early detection of three of the cancers are presented.