A radiochemical oestrogen receptor assay on cytosol was correlated with a radiochemical and an immunohistochemical oestrogen receptor assay using cryostat sections from 50 breast cancer specimens. Oestrogen receptors were reliably quantitated in 6 micron cryostat sections with Scatchard analysis using radiolabelled oestradiol, and a good quantitative and qualitative relation with cytosol oestrogen receptor assay was found. Parallel sections were used for routine histological tissue verification and for direct comparison with immunohistochemistry for oestrogen receptor. Specific immunoperoxidase staining with a rat monoclonal antibody was scored by semiquantitative evaluation of the staining intensity of cancer cell nuclei. Oestrogen receptor scoring was highly reproducible when performed by the same observer. The semiquantitative immunohistochemical oestrogen receptor score correlated significantly better with the radiochemical assay on sections than with cytosol assay. Oestrogen receptor in breast cancer can be reliably assayed by semiquantitative evaluation of cryostat sections immunostained for oestrogen receptor, but only if the procedure is adequately standardised. The results underline the importance of cellular heterogeneity as a cause of variation in oestrogen receptor assay evaluation in breast cancer.