Metaplastic breast carcinomas are rare neoplasms showing both carcinomatous and sarcomatous elements. In this report we describe eleven cases of metaplastic breast carcinoma focusing on pathological features and the clinical behaviour of six patients with breast carcinoma with chondroid metaplasia (MCC). We collected eleven cases from 1996 to 2001: immunohistochemical tests were performed in order to obtain data on estrogen and progesterone receptors and the production of p53 gene and HER/2 neu. Neoangiogenesis was studied counting vessels immunohistochemically-stained with CD31 antibody. Six cases showed chondroid metaplasia, three cases were spindle cell carcinoma and two were metaplastic squamous carcinoma. The majority of patients (64%) had pT2 tumors without axillary node metastases: only two cases with spindle or squamous metaplasia showed nodal involvement. Fifty percent of MCC were pT1b-c tumors: no axillary metastases were observed. Vascular invasion was observed in all squamous and spindle cell types and in 66% of MCC: estrogen and progesterone receptors were absent in 90% of the tumors. Immunohistochemical staining for HER2/neu was detected in 72% of spindle cell and squamous carcinomas and in 33% of MCC. Three cases staining highly for p53 were chondroid carcinomas: the staining was uniform both in carcinomatous and in sarcomatous tissue. The majority of metaplastic carcinomas had high angionesis. One patient with a chondroid metaplastic carcinoma was found to be a carrier of a BRCA1 mutation similar to the one responsible for sickle cell disease, possibly altering the spatial structure of the gene product. Only six patients had follow-up periods longer than 36 months: five women were alive and disease-free: one patient with pT2N1 squamous metaplastic carcinoma died of disease 14 months after diagnosis. The six women with MCC were alive and disease-free. Surgical and adjuvant treatment should follow the guidelines for the other most common breast cancers even if the need for chemotherapy is unknown due to the absence of large series randomized or observational data.