Micrometastases of solid tumors are most commonly detected by immunocytochemistry using monoclonal antibodies directed against tissue-specific gene products like cytokeratin-18 (CK-18) and the carcinoembryonic antigen (CEA). While CK-18 is a marker for epithelia in general, CEA is mainly employed in the detection of gastrointestinal and breast carcinomas. To improve the sensitivity and specificity of micrometastasis detection, we planned to establish polymerase chain reaction (PCR) assays for both markers. Here we provide strong evidence for the existence of a CK-18 pseudogene, since specific amplification (i) was readily obtained from healthy bone marrow donors, (ii) did not require reverse transcription of CK-18 mRNA and (iii) was not abolished by RNase treatment. Using a CK-18-specific probe, Southern blot analyses revealed identical-size fragments for both genomic DNA and a CK-18 cDNA after digestion with appropriate restriction enzymes. On the other hand, the amplification of CEA mRNA (i) was never observed in bone marrow samples of healthy donors or patients without solid tumors, (ii) required intact mRNA and the reverse transcriptase reaction, and (iii) could not be obtained after RNase treatment. In reconstitution experiments, single CEA-expressing tumor cells were reliably detected among 2 x 10(7) normal bone marrow cells. We conclude that, due to the presence of pseudogene(s), PCR-based detection systems are not readily suitable for CK-18, while the CEA mRNA amplification should provide a sensitive and specific test for the presence of ectopic, and hence presumed malignant, CEA-expressing cells in body fluids.