Surgical biopsy specimens of 179 breast carcinoma were studied by steroid-binding and immunohistochemical assays or oestrogen and progesterone receptors (ER, PR) in order to explore reasons for discordant results between the two assay types. Receptor statuses in 18% of ER assays and 30% of PR assays were in disagreement. Immunohistochemistry-positive steroid-binding-negative status predominated among the discordant ER assays, while the discordant PR assays displayed the opposite situation. In discordant assays receptor concentration was significantly more often close to the cut-off (10-50 fmol mg-1) than in the concordant ones. Low binding affinity (high Kd) was also significantly associated with disagreeing assay results. These observations clearly indicate that immunohistochemical ER and PR assays measure high-affinity binding components (i.e. type I receptors) in steroid-binding assays. ER but not PR assays in premenopausal women disagreed more often than those in post-menopausal women. Such factors as histological type, specimen size in steroid-binding assay, grade of malignancy and tumour necrosis were statistically unrelated to agreement or disagreement of receptor assays.