Breast cancer patients (n = 224) aged 28 to 81 were postoperatively followed up with serial determinations of carcinoembryonic antigen (CEA), tissue polypeptide antigen (TPA), and urinary hydroxyproline (OHP). The clinical usefulness of these tumor markers to diagnose and monitor distant metastases was compared with that of the imaging techniques commonly used to monitor breast cancer patients (bone scanning [BS], liver echography [LE], chest radiograph, and skeletal radiograph). So far, 23 patients withdrew from the study, and distant metastases occurred in 33 patients. In 91% of the metastatic patients, constant elevation or progressive increase in serum CEA and/or TPA levels were the first pathologic findings of the relapse. Of the remaining 168 nonrelapsed patients, 122 were followed up longer than 24 months (43 +/- 17 months; mean +/- SD). In these 122 patients the false-positive results of CEA, TPA, and OHP were 0.8%, 2.4%, and 0%, respectively, when used simultaneously with clinical examination and the common laboratory examinations. BS and LE are the only imaging techniques that showed such a high sensitivity to be suitable in the postoperative follow-up of breast cancer patients. Nevertheless, because BS has a low specificity and is not harmless, it should be performed at longer intervals than tumor markers. Eventually, in the relapsed patients, TPA and OHP well reflected the response to treatment better than CEA and prevented useless radiologic examinations.