Epidemiological and observational studies suggest that oestrogens, when used as hormonal therapy in post-menopausal women, can increase the risk of breast cancer if used long term. However, more recent data suggest that short-term use in sub-groups of post-menopausal women significantly decreases the risk of breast cancer. This beneficial effect is also observed when high-dose oestrogen is administered to post-menopausal women with breast cancer to cause tumour regression, a phenomenon which commonly occurs. We consider these divergent responses to oestrogen to represent a "paradox". Data from our own and other investigative groups suggest a hypothesis to explain this paradox. Deprivation of oestradiol in model systems causes cells to adapt and to undergo apoptosis in response to oestrogen. This occurs through the Fas/Fas ligand death receptor pathway and through alterations in apoptotic mechanisms mediated by mitochondria. This process of programmed cell death may explain the regression of established breast cancer with oestrogen administration and the diminution in the rate of new breast cancer diagnoses recently reported. Our hypothesis is based upon pathological data indicating the presence of a "reservoir" of undiagnosed breast cancer in the population of women who would be starting on oestrogens as menopausal hormonal therapy. The long-term increased risk of breast cancer may then reflect different mechanisms. Oestrogens can cause mutations through enhancement of the rate of cell division and concomitantly the error rate in DNA replication. In addition, oestrogens can be metabolised to directly genotoxic compounds. These carcinogenic processes take much longer, since a number of mutations must accumulate before resulting in breast cancer. These hypotheses regarding oestrogen-induced apoptosis in the short term and carcinogenesis in the long term now require rigorous verification but would serve to explain the "oestrogen paradox". (Pol J Endocrinol 2007; 58 (3): 222-227)