Most human invasive breast cancers (IBCs) appear to develop over long periods of time from certain pre-existing benign lesions. Of the many types of benign lesions in the human breast, only a few appear to have significant premalignant potential. The best characterized of these include atypical hyperplasias and in situ carcinomas and both categories are probably well on along the evolutionary pathway to IBC. Very little is known about earlier premalignant alterations. All types of premalignant breast lesions are relatively common but only a small proportion appear to progress to IBC. They are currently defined by their histological features and their prognosis is imprecisely estimated from indirect epidemiological evidence. Although lesions within specific categories look alike, they must possess underlying biological differences causing some to remain stable and others to progress. Recent studies suggest that they evolve by highly diverse genetic mechanisms and research into these altered pathways may identify specific early defects that can be targeted to prevent premalignant lesions from developing or becoming cancerous. It is far more rational to think that breast cancer can be prevented than cured once it has developed fully. This review discusses histological models of human premalignant breast disease that provide the framework for scientific investigations into the biological alterations behind them and examples of specific biological alterations that appear to be particularly important.