The association between estrogen replacement therapy and female breast cancer was studied in two Los Angeles area retirement communities. The 138 study cases of breast cancer occurring in residents younger than 75 years were compared with age- and race-matched community control subjects. The risk ratio for a total cumulative dose in excess of 1,500 mg was estimated to be 2.5 in women with intact ovaries. This increase was present using various independent sources of drug usage information but was inconsistent at low dose and undetectable in oophorectomized women. No important sources of confounding could be identified, and no risk modifiers could be identified except for a history of surgically confirmed benign breast disease. In such women with intact ovaries, the risk ratio for a high cumulative dose rose to 5.7 relative to nonusers with normal breasts.