Recent studies have shown that chromosome 8p21-22 is the main site of frequent loss of heterozygosity (LOH) in breast cancers. However, the detailed molecular analysis of chromosome 8 so far in breast cancer has been variable. Most of the literature pertaining to LOH in breast cancer is mainly on short arm of chromosome 8. In the present study, we have examined LOH on both short and long arm of chromosome 8 using fifteen different polymorphic DNA markers in microdissected samples of normal breast epithelium and carcinoma from the same patients. For this purpose, DNA was extracted from the microdissected normal and tumor cells of 66 breast cancers, amplified by PCR and analyzed for LOH on chromosome 8 using fifteen different polymorphic DNA markers (D8S264, D8S298, D8S535, D8S255, D8S1098, D8S589, D8S567, D8S591, D8S285, D8S1102, D8S1763, D8S260, D8S530, D8S1772, and D8S1844). Expression of estrogen receptor, progesterone receptors, and p53 antigens was determined by immunohistochemistry using specific monoclonal antibodies. The results of this study suggest that LOH on chromosome 8 was identified in 40 of 66 cases (61%) with at least one marker. Three distinct regions of loss detected were: i) at 8p12, at loci between D8S535 and D8S255; ii) at 8p11, on loci D8S567, D8S591, D8S285, and D8S1102; iii) at 8q11-12, on loci D8S1763, D8S260 and D8S530. We found 45% (30 out of 66 informative cases) of the tumors showed LOH at 8p12; 52% (34 out of 66 informative cases) had LOH at 8p11; and 39% (26 out of 66 informative cases) had LOH at 8q11-12. Deletion at 8q11-12 was significantly correlated with the grade of the breast cancer specimens. Moderate to poorly differentiated specimens had higher incidence of LOH at 8q11-12 as compared to well differentiated specimens. Deletion at 8p12 and 8p11 was significantly higher in clinical stages III and IV of breast cancer tissues as compared to stage I and II cases. Tissues with lymph node involvement showed higher incidence of LOH at 8p12 as compared to the tissues with no lymph node involvement. There was no correlation of LOH at these loci with either the age of the patients, tumor size, BrdU labeling index, expression of estrogen receptor, progesterone receptor, and p53 in breast cancer specimens. These experiments, for the first time, report multiple sites of LOH on chromosome 8 in human breast cancer, and these deletions have differential correlation with clinical parameters of breast cancer samples.