An enzyme immunoassay was compared with a radioligand assay (isoelectric focusing) for estrogen receptor (ER) determination in 100 preoperative fine-needle aspirates from human breast cancer. A strong correlation (rs = 0.85) between the two estimates was found. However, ER estimates by enzyme immunoassay were 6 times higher than with isoelectric focusing, mainly due to an underestimation of ER content with the latter method in highly blood-contaminated or diluted samples. Enzyme immunoassay provided adequate ER measurement even at very high (50%) blood admixture. To study the reliability of ER measurement with enzyme immunoassay in fine-needle aspirates, a comparison was made with estimates in the corresponding surgical biopsies (65 cases). With DNA as reference parameter a significant correlation (rs = 0.68) was obtained. Excluding aspirates with low cellularity (less than 50,000 cells/ml), only one ER negative fine-needle aspirate was found among 47 ER positive surgical biopsies. A few ER positive aspirates corresponded to ER negative surgical biopsies possibly due to tumour heterogeneity or the fact that fine-needle aspiration sometimes may represent a better sampling technique than surgical biopsy.