There is general consensus that screening can reduce mortality from colorectal, breast, and cervical cancer among persons in their 50s and 60s. However, few screening trials have included persons over age 70 years. Therefore, indirect evidence must be used to determine when results in younger persons should be extrapolated to older persons. In this review, we focus on cancer screening tests that are well accepted in younger persons (mammography, Papanicolaou smears, and colorectal cancer screening) and discuss the strength of inference concerning benefits and harms of screening older persons. Some aspects of aging favor screening (eg, increased absolute risk of dying of cancer) whereas other aspects do not (eg, decreased life expectancy). Age also affects the behavior of some cancers (eg, increases the proportion of slow-growing breast cancers) and affects the accuracy of some screening tests (eg, increases the accuracy of mammography; decreases the accuracy of sigmoidoscopy). These effects make the application of evidence in younger populations to older populations complex. However, given the heterogeneity of the elderly population, there is no evidence of one age at which potential benefits of screening suddenly cease or potential harms suddenly become substantial for everyone. Therefore, characteristics of individual patients that go beyond age should be the driving factors in screening decisions. For example, persons who have a life expectancy less than 5 years or persons who would decline treatment should generally not be screened. Decisions to either continue or discontinue screening in the elderly should be based on health status, the benefits and harms of the test, and preferences of the patient, rather than solely on the age of the patient.