Frozen section diagnosis rendered in 549 consecutive breast biopsies performed in 5 years in a single pathology laboratory was correlated with the final pathological diagnosis. There were no false positive reports among the 220 (40.1%) biopsies interpreted as benign lesions in paraffin sections. Among 329 (59.9%) malignant biopsies on paraffin sections, 3 cases were interpreted as benign lesions on frozen sections. Three false negatives included 2 ductal carcinoma in situ and one infiltrating ductal carcinoma associated with papillomatosis. The tumors were small and confined to the breast without any evidence of metastasis. There was a very good correspondence between the frozen section diagnosis and the paraffin section diagnosis (K = 0.98). The sensitivity of frozen section diagnosis was 99.1% and the clinical diagnostic specificity was 100%. Our results suggest that frozen section diagnosis is a highly reliable procedure, but small lesions (less than 1 cm in diameter, or non-palpable) should not be subjected to frozen section examination to avoid unnecessary loss of neoplastic tissue during the frozen section. The careful investigation of paraffin-embedded tissue is recommended for small breast lesions in breast conserving lumpectomy.