The 3-step, unlabeled antibody, immunoperoxidase method for staining CEA in conventional histopathology sections was used to evaluate the presence of CEA in 950 different specimens, the vast majority of which were available as paraffin-embedded tissues fixed in formalin. The sensitivity of the method clearly discriminated between normal or nonmalignant, diseased tissues and neoplasms having increased quantities of CEA. Thus, tumors of epithelial origin, particularly adenocarcinomas or squamous cell carcinomas of the gastrointestinal tract, bronchus, ovary, and cervix, were frequently prone to CEA staining. Among ovarian tumors, there appeared to be a distinct proclivity of CEA staining in mucinous as compared to serous cystadenocarcinomas. The only normal tissues showing CEA were colonic mucosae adjacent or near to benign or malignant tumors, suggesting CEA absorption from the neoplasm and not primarily de novo CEA biosynthesis. With the exception of inflammatory conditions of the colon, nonmalignant, diseased tissues did not have CEA demonstrable by this method. It is proposed that immunocytochemical staining of CEA in conventional tumor pathology sections can be used to discriminate those cases in which blood CEA titers need to be determined on a serial basis in order to monitor recurrence or progression, since a positive immunoperoxidase reaction for CEA in the primary tumor predicts well in many cases which patients will have a blood CEA value reflecting disease activity. Furthermore, the excellent correlation found for CEA staining between primary and metastatic tumors of the same cases suggests that the immunocytochemical method could have diagnostic potential for detecting micrometastatic spread of CEApositive cancers to regional lymph nodes, thus aiding in the screening of regional lymph nodes of cancer patients.