A review of assay results from more than 6000 patients revealed 232 patients in whom multiple breast cancer specimens were analyzed for estrogen receptor (ER). All assays were performed in a single laboratory. Specimens were considered estrogen receptor positive (ER+) if the ER level was greater than 10 fmol/mg protein and estrogen receptor negative (ER-) if the ER level was less than 3. ER values between 3 and 10 fmol/mg protein were considered borderline. Simultaneous assays were performed in 58 patients with 3% major discordance (i.e., one assay ER- and one assay ER+). Major discordance for sequential biopsies was 19% (16 of 82) when the initial assay was ER+ and 13% (eight of 63) when the initial assay was ER-. (Apparent change from ER- to ER+ status was observed in five of nine patients with primary tumors less than 2 cm in diameter, suggesting that an inadequate amount of tissue may have been submitted for initial ER analysis.) There was no significant relationship between the time interval between sequential biopsies and the rate of discordance. Marked decreases in ER levels and 78% discordance were seen if rebiopsy was performed within 2 months of tamoxifen treatment. When these tamoxifen cases were excluded from the analysis, neither intervening endocrine therapy nor chemotherapy significantly altered discordance rates.