We have studied estrogen receptor (ER) and progesterone receptor (PR) in normal breast by immunocytochemistry using tissue biopsies and fine needle aspirates (FNA) and, in the case of ER, by enzyme immunoassay. For ER we found a high degree of reproducibility for biopsies taken from the upper outer quadrant: FNA, r = 0.56 (P less than 0.002); tissue section immunocytochemistry, r = 0.89 (P less than 0.0001); and enzyme immunoassay, r = 0.76 (P less than 0.0001). For PR, FNA (r = 0.56, P less than 0.002) and tissue section (r = 0.97, P less than 0.0001) were also found to be reproducible techniques. Using enzyme immunoassay, we were able to measure ER accurately in normal breast tissue. In 59 samples we found a range of 0-37 fmol/mg cytosol protein (mean, 4 fmol/mg). In an age-matched group of 126 women with breast cancer, we found a significantly higher ER [range, 0-139 fmol/mg; mean, 37 fmol/mg (P less than 0.001)]. We then analyzed the ER and PR content of FNAs obtained from the upper outer quadrant of the normal breast in 143 normal women. We found that in only 23 of 143 samples (16%) were greater than or equal to 50% epithelial cells stained. There was a relationship between ER and PR (P = 0.03) and a higher ER content in European women than in non-European women (P less than 0.03). The PR content was related to high body mass index (P less than 0.02) and family history of breast cancer (P = 0.04). Samples tended to be more frequently ER positive by FNA if taken in the follicular phase of the menstrual cycle. We conclude that, although the levels of ER and PR are low in normal breast, they can be accurately measured. There is significant variation of ER and PR with several clinical parameters.