This review focuses on the effects of oral contraceptives (OC) and estrogen replacement therapy (ERT) on the risk of breast, ovarian and endometrial cancer. The relationship between OC and cancer risk is first placed in a historical perspective. Since 1960, when OC were introduced, the hormonal composition of OC as well as the characteristics of the OC user have changed considerably. Studies conducted in the 1970s were generally reassuring, but it was not until the 1980s that studies could evaluate the effect of prolonged OC use after an extended follow-up period. Although the relationship between breast cancer and OC has been investigated in about 40 studies, the issue still remains essentially unresolved. Most studies report no association between ever use of OC and breast cancer risk. Several studies find increased risk for prolonged use and other studies report elevated risks for women who used OC very early in their reproductive years. The inconsistent results of recent studies are attributed to bias or to geographical variation in latency period elapsed, types of OC preparations, or prevalence of other risk factors. In contrast, the use of combined OC has consistently been shown to reduce the risk of ovarian and endometrial cancer. The risk further decreases with increasing duration of use and the protective effect seems to persist in ex-users for at least 5 years. Some evidence indicates that higher parity reduces the protective effect. Though studies relating ERT to breast cancer are far from consistent, overall, there is evidence for a moderately increased risk with high dose and/or long duration. The effect seems to be modified by mode of administration (injections vs. pills) and by type of ERT, but this needs confirmation. The number of adequate studies on the relationship between ERT and ovarian cancer is too small to draw firm conclusions. The positive relationship between ERT and endometrial cancer is now well established. The ERT effect is dose- and duration-dependent and is characterized by a short latency period. The cyclic addition of progesterone (greater than 10 days/cycle) may reduce the risk increase.