In 57 patients with metastasized breast carcinoma, serum CEA levels (CEA-Ria-Kit, Abbot) were correlated with the clinical course under endocrine or cytostatic therapy. In 20 patients, CEA was determined at intervals of 1 to 3 days over the first 4 weeks of treatment and in 21 patients at intervals of 1-3 weeks. In 16 patients, CEA was determined prior to treatment. The clinical development was judged according to the program of UICC. We found pretreatment CEA values higher than 25 ng/ml serum mostly in patients with liver involvement and in patients with progression under the following therapy. In 22 patients, progress or remission of the disease were correlated with CEA increase or decrease of more than 20% during the first 8 weeks. In 11 patients, we found no correlation between CEA and clinical course. From the remaining 8 cases, CEA levels were out of the measuring range in 6 patients, in 2 patients we were unable to interpret the clinical course doubtlessly. There were no short-termed CEA alternations after the start of the treatment rendering a possible early estimation of the therapy effect. In 5 of 23 patients treated with methylprogesteronacetate and aminoglutethimide a continuous CEA increase developed despite clinical remission. This was not found in 10 patients treated with cytostatic chemotherapy. The pretreatment CEA is a useful prognostic parameter in patients with metastasizing breast cancer. CEA should be determined weekly in patients with increasing CEA levels under therapy and monthly in patients with decreasing CEA levels. The possibility of a tumor-independent CEA rise as a side-effect of treatment must be taken into consideration.