A review of the tumor biology of breast cancer from the screening perspective reveals important research issues. It is not known if the induction phase includes any detectable preneoplastic lesions. Evidence for the existence of preneoplasias is conflicting. The effect of removal of such lesions on the incidence of breast cancer has not yet been studied. The mechanisms that govern progression of in situ lesions--or even small preclinical invasive cancers--are not fully understood. It is not clear to what extent progression into cancers with metastatic behavior occurs. Current predictors for progression are weak. One major benefit of the scientific evaluation of the screening programs is that we can learn more about the natural history of the disease. Mammographic screening detects tumors that are smaller in size than those detected by clinical examination only. There is an inverse relationship between tumor size and the probability that the patient has acquired axillary metastases. These facts, taken together, indicate that the mortality reduction seen in screening is a result of a real impact on the natural history. Calculations of the lead time gained in randomized screening projects give empirical proof that a third of the cancers progress from a localized stage to a disseminated disease relatively late in the preclinical phase of the tumor. Analyses of interval cancers in screening have pointed to the conclusion that the net growth rate of the primary tumor does not directly parallel metastatic ability, and that the tumors grow faster in younger women.