From April 1987 to December 1990 490 women and 5 men suffering from primary or metastatic breast cancer underwent 1485 assays of the CA 15-3 marker using a radioimmunometric method supplied by Centocor based on two monoclonal antibodies: 115D8 and DF3. The dose-response curve constructed using 5 standards. Clinical and instrumental monitoring of patients continued for at least one year after the last assay. In line with the bibliographical data and the Author's extensive experience a cut-off 40 U/ml was selected. On this basis the percentages of sensitivity (58.2), specificity (96.0), accuracy (85.0), and positive (85.7) and negative (84.9) predictive value were calculated and were in overall terms better than those obtained using a threshold of 30 U/ml. A chart marking the trend of percentages for assays in patient with active metastasis showed that there was a rapid increase over 40 U/ml, reaching 100% over 90 U/ml. In asymptomatic patients, who were considered "ned" following clinical and instrumental controls, it was rare to find and assay level over 50 U/ml (13/1050). The authors underline the importance of serial controls in post-treatment follow-up at intervals of not less than one month and not more than 3 months for 1-2 years, and subsequently every six months. This marker has been found to offer considerable prognostic, but not diagnostic, reliability during the preoperative period and even more so during cancer control due to the possible early detection of metastasis in the viscera or skeleton, especially if this is conspicuous. On the other hand, the marker shows limited sensitivity to cutaneous or regional lymph node metastasis. In this pathology levels in excess of 200 U/ml have never been observed. CA 15-3 is particularly important in evaluating the efficacy of anti-tumour therapies or deciding whether they are worth continuing. In conclusion, a few general comments are made regarding the indications for doses and the value of the marker in controlling patients treated for breast cancer.