Homogenates of breast tumours taken at surgery were prepared in phosphate-buffered medium in the presence or absence of the protease inhibitors N-alpha-p-tosyl-L-arginine methyl ester (TAME, 10 mM) or soy bean trypsin inhibitor (STI, 1 mg/ml). Aliquots (0.25 ml) were incubated in 5 ml medium, with the addition of excess trypsin (2 mg/ml) to experimental flasks. Oestrogen was measured by means of a radioreceptor assay (RRA) based on rat or human uterine cytosolic oestradiol receptor. In oestrogen receptor positive (ER +ve) tumour homogenates, TAME decreased while STI increased ethyl acetate extractable oestrogen in these preparations. The addition of trypsin enhanced yields of oestrogen in the TAME, but not in the STI or control (no inhibitor) preparations. None of these treatments affected RRA detectable oestrogen in homogenates of ER -ve tumours. Suspensions of ZR-75-1 cells, prepared in Krebs Ringer bicarbonate (KRBG) incubated with trypsin also gave greatly enhanced yields of extractable oestrogen. Fractionation of oestrogens from both tumour homogenates and from the cell line showed coincidence of RRA detectable steroid with oestradiol and oestrone, and, particularly in trypsin flasks, very non-polar components were also found. In the cell-line extracts, HPLC fractionation combined with specific radioimmunoassays confirmed the presence of both oestradiol and oestrone. The major extracted component was oestrone. The data suggest the existence within breast tumour tissue of sequestered pools of steroid requiring proteolytic action for their release. One possibility, consistent with reports in the literature, is that the steroids may themselves be directly conjugated to protein. Their presence in ER +ve but not ER -ve tumours strongly suggests some relationship to the development of hormone-sensitive disease. Alternatively, the phenomenon may be associated with the rigid compartmentalization of the paracrine function of the tissue.