Serum carbohydrate antigen 15.3 (CA 15.3) and carcinoembryonic antigen (CEA) are currently employed in clinical practice as markers for breast cancer, particularly in the follow-up and therapy monitoring. However, the American Society for Clinical Oncology (ASCO) stated in its clinical practice guidelines for the use of tumour markers in breast carcinoma that neither CA 15.3 nor CEA are recommended for routine use in screening, diagnosis and surveillance after primary treatment, or in monitoring response to treatment, because current literature data are insufficient. Cytokeratin fragment 21.1 (CYFRA 21.1) assay detects a serum fragment of cytokeratin 19 (CK19) and is employed in the diagnosis and management of lung cancer, particularly of squamous cell histotype. Breast carcinoma has been demonstrated to express CK19 fragments in the primary and metastatic lesions and CK19 mRNA is detectable in peripheral blood from patients affected by breast cancer. We measured serum markers CYFRA 21.1, CEA and CA 15.3 in the sera from 212 females affected by histologically proven breast carcinoma. Patients comprised 96 individuals with untreated primary disease (54 stage I-II, 18 stage III and 24 stage IV), 30 regional (chest-wall and/or lymph-nodes) relapsing disease and 68 metastatic (haematogenous metastases) relapsing disease. Forty-eight patients previously treated by surgery and without any evidence of disease were enrolled to evaluate the role of serum markers in the monitoring for recurrence of the disease. One hundred healthy age-matched females and 65 patients affected by benign mammary gland disease (including 38 patients with mastopathy and 27 with fibroadenoma) were enrolled as controls. Serum levels of all markers increased from controls to patients affected by breast cancer, from stage I-II to stage IV of the breast cancer and from local to advanced recurrence. The comparison of diagnostic accuracy in the detection of primary and relapsing breast cancer showed no significant differences between markers. Univariate and multivariate survival analysis showed a significant statistically prognostic value for CA 15.3 and CYFRA 21.1 but not for CEA. However, the factors N and M were confirmed to be very strong predictors of the patients' survival. Finally, CEA and CYFRA 21.1 detected less recurrences than CA 15.3. In conclusion, our data show no significant improvement in the diagnosis, prognostic evaluationand follow-up of breast cancer by CYFRA 21.1 and CEA assays compared to CA 15.3 assay. Considering the ASCO statement on tumour markers in breast cancer, the CYFRA 21.1 assay should not be employed in clinical practice.