Summary Assessment of heterogeneity in oestrogen receptor (ER) expression aims to improve prediction of prognosis and treatment assignment in breast cancer. Current assessments are performed manually and are subjective. Automated image analysis as described here objectively quantitates ER in breast cancer nuclei obtained by needle aspiration. ER was visualised by ERICA with diaminobenzidine (DAB) substrate. Various indices of ER positivity were derived from the integrated density and average density measurements of nuclear DAB. Each index was compensated for background staining by non-specific antibody binding and endogenous peroxidase activity. Total nuclear ER content (iAntegrated optical density of stain) was strongly associated with the biopsy ER concentration determined by shturation analysis of radioligand binding (DCC), P < 0.005. Nuclear ER concentration by image analysis (mean optical density of stain) was The presence of oestrogen and progesterone receptors in breast cancer tissue is now accepted as indicating a better overall disease prognosis for the patient, and as a marker for assisting the selection of patients who may benefit from endocrine therapies (Cant et al., 1985;McGuire, 1987;Thorpe et al., 1986;Williams et al., 1987).However, the power of prediction for good clinical endocrine response which is afforded by hormone receptor quantitation could be improved considerably at the individual patient level. The current feeling is that a knowledge of the heterogeneity of nuclear oestrogen receptor (ER) expression could be helpful in improving the reliability of patient assessment, and thereby assist in the assignment of appropriate treatment. Fine needle aspiration (FNA) biopsy is a proven effective procedure for the diagnosis of malignant diseases of the human breast , and a simultaneous evaluation of FN aspirates for the heterogeneity of oestrogen receptor expression within the cancer cell population would yield an early assessment of prognosis without recourse to surgery. The newly emerging technique of video image analysis promises to remove much of the subjectivity integral to semi-quantitative assessments of immunocytochemical staining (Charpin et al., 1986).Our aims in performing this study were: (1) Video image analysis of ERICA stained aspirates The video image analytical system used in this study was developed in our laboratories (Jarvis, 1986(Jarvis, , 1987(Jarvis, , 1988. In brief, the system comprises a microscope, solid state camera, video digitiser, microcomputer and digitiser tablet (Jarvis, 1988). The image analysis software permits selection and scoring of individual tumour cells through editing of debris, overlapping or touching nuclei, stromal cells or any other undesired feature. The editing functions are performed by a Br.