The use of different techniques for assay of oestrogen receptors (ER) in breast cancer raises the question of their relative effectiveness in measuring concentrations of functional receptors. Data were obtained on soluble receptors from supernatants from 58 primary breast tumour homogenates, using the ligand ([3H]oestradiol) binding assay with dextran-coated charcoal (DCC) separation, either at a single saturating ligand dose, or by Scatchard analysis, and by using the Abbott enzyme immunoassay (EIA) kit. As previous reports have shown, the two methods gave reasonably good correlation (r = 0.8), but EIA values were systematically higher than DCC (slope = 3.0). Similar values were obtained when the ER + ve/progesterone receptor (PR) + ve subgroup were examined separately (n = 34, r = 0.86, slope = 3.0). However the two sets of data were in much better agreement in the ER + ve/PR - ve subgroup (n = 10, r = 0.98, slope = 1.24). When analysed by isoelectric focusing on polyacrylamide gels (IEF), two major specific binding components were identified, at pI 6.1 and at pI 6.6. Both isoforms were present in 50/66 ER + ve PR + ve breast tumour samples, but only the pI 6.6 (4S) was present in most ER + ve/PR - ve samples (13/20). It appears that, compared with DCC, the EIA method gives much higher values for the 8S isoform, whereas the two methods detect the 4S isoform with similar sensitivity. In assays on the tumour cell lines, T47D and MCF-7, still greater discrepancies, at least 10-fold, were found between EIA and DCC data.