Shedding of neoplastic cells into the circulation is an essential event for the hematogenous metastasis of solid tumors. Recently, several studies reported that a high frequency of cancer cells could be detected in the bloodstream during surgery. The intraoperative detection of hematogenous dissemination of cancer cells was able to identify a subset of patients with malignant diseases at high risk for postoperative metastasis and to predict a poor prognosis. In order to evaluate the association between intraoperative dissemination of cancer cells and postsurgical survival of patients with non-small cell lung cancer (NSCLC), we developed a flow cytometric assay for specific detection of lung cancer cells in the blood. The monocyte-enriched population in the blood was separated by a modified Ficoll-Hypaque density centrifugation and then labeled with a combination of monoclonal antibodies specific for CD45, cytokeratin (CK) and two antigens expressed on lung cancer cells (2F7 and S5A). The assay could detect quantitatively lung cancer cells (defined as CD45(-1) CK(+) 2F7/S5A(+) cells), with the sensitivity limit of one cancer cell in 10(5) normal leukocytes. The specificity for lung cancer was 97%, which was calculated from the results of healthy subjects (20 cases) and patients affected with benign pulmonary diseases (26 cases) or esophageal cancer (14 cases). Blood samples of 31 NSCLC patients were collected from pulmonary vein during open thoracic surgery. Fifteen of them (48.4%) were found to have positive test results. The average cancer cell counts in these cases were 0.306 x 10(6)/l. Patients under 55 years of age had a significantly higher percentage of positive findings than those over 55 years of age (P < 0.05). The positive rate increased over the stages and lymph node status, but the differences were not statistically significant. Moreover, patients with squamous cell carcinoma at later stages (stages III and IV) had an increased frequency of positive test results than those at earlier stages (stages I and II, P < 0.05). In contrast, no such a difference was found in cases with adenocarcinoma. On the basis of 30-months follow-up date, the median survival time and 2-year survival rate for patients with positive and negative findings were 11 vs. 27 months, and 26.7 vs. 62.5%, respectively. There was a statistically significant difference between overall survival curves that favored the patients with negative test results (P = 0.023). Multivariate analysis indicated the stage of disease and the positive test results as two independent factors that affected survival time (P = 0.017 and 0.027). When a comparison was made within the patients at stages III and IV, the presence of cancer cells in blood was associated with a significantly shorter survival. These data indicate that the hematogenous dissemination of lung cancer cells during surgery would be one of the mechanisms of postoperative tumor metastasis. The detection of these cells may help to identify patients with poor prognosis.