Abstract
The evaluation of screening programs for cancer is considered. Initial evaluation has to be in terms of mortality, but increasing importance should be attached to measures which evaluate the interaction of early detection with the disease process. These include the degree to which diagnosis is advanced, both in time and in stage of disease, and the ability of the screening test to identify presymptomatic lesions. Randomized trials are of major importance in establishing—in quantitative terms—the benefits of screening; in their absence, case‐control methods can be adapted for this purpose, but cannot provide evidence of the same rigor. Currently, only screening for breast cancer and cervical cancer have been fully demonstrated to be effective.