Recently, there has been significant effort in developing techniques designed to detect disseminated tumor cells in the peripheral blood (PB). These techniques include immunocytochemical staining of cytocentrifuge slides, flow cytometry, and RT-PCR. Several authors reported various results concerning the sensitivity of the detection limit when applying these methods. The aim of this study was to assess the value of two methods in the detection of ovarian carcinoma cells in the PB. For tumor cell detection we compared RT-PCR to immunomagnetic enrichment followed by flow cytometric analysis. In a model system, single cell suspensions of ovarian cancer cell lines were mixed with full blood samples from healthy donors in order to determine the sensitivity limit of the two methods and to analyze the reproducibility of each. In a multiparameter flow cytometric analysis, tumor cells were defined as cytokeratin 7/8 positive and CD45 negative. RNA was screened for MUC1 mRNA by RT-PCR. MUC1 mRNA expression turned out not to be a specific marker of disseminated ovarian cancer cells, because a weak expression was also found in samples of healthy persons. Using immunomagnetic enrichment followed by flow cytometry, one carcinoma cell per 1 x 10(6) leukocytes was detectable. However, a minimum of 10 ml blood had to be analyzed in order to clearly distinguish real positive tumor cells from false-positive signals.