BACKGROUND: Popular immunohistochemical techniques for assay of estrogen receptor(ER) allow the localization of positive cells in specific cell populations. Some of breast carcinomas composed of discrete populations of cells were negative for ER, while neighboring populations of cells were positive for ER. Such heterogeneity might be due to biological or artifactual causes. METHODS: We studied 67 tissue blocks for geographic heterogeneity within the level of ER and cytokeratin(CK) by staining ER and CK. Positive distribution of ER and CK was manually assessed. RESULTS: The immunohistochemical expression revealed 50 cases for ER-positive and 17 cases for ER-negative. In 50 ER-positive cancers, homogeneity was 38 cases, heterogeneity was 11 case, and artifactural change was developed in and one case. excluded in the analysis. The rate of heterogeneity of the ER-positive cancers was 22.4%(11/49). Comparisons of homogeneity and heterogeneity according to clinicopathologic risk factors in ER-positive breast cancer demonstrated that the heterogeneity of ER was significantly higher in each subgroups; relatively younger ages(< or =50yr), premenopausal status, early menarche(< or =15yr), early stage(< or =I b), DCIS in pathology, and lower positive expression rate of ER(<50%). CONCLUSION: Clinicopathologic risk factoes would be required to discover the heterogeneity of ER-positive breast cancer. Also a long-term follow-up study on risk factors, including disease free survival, response to anti-estrogen therapy, and survival according to heterogeneity of ER would be need.