BACKGROUND: To enhance the cancer cell detection rate in blood, we tried to detect cancer cells by blood filtration, RNA extraction and reverse transcription (RT)-PCR in the filtered mononuclear cells (MNCs). METHODS: From the specimens of whole blood and filtered MNCs, RNA was extracted by the guanidinium isothiocyanate buffer method. Filtration efficiency was evaluated by measurement of leukocyte count, red cell count, and hemoglobin level. To compare the RNA extraction efficiency between whole blood and filtered MNCs, the followings were examined: (1) RNA electrophoresis, (2) hTERT, survivin, plakophilin, LunX, and MAGE A1-6 RT-PCR, and (3) the detection limit of added SNU484 cells in the blood by MAGE A1-6 RT-PCR. Finally blood specimens of 13 lung cancer patients were used to detect cancer cells by LunX and MAGE A1-6 RT-PCR with filtered MNCs. RESULTS: The filtration method revealed 0%, 92.8% and 95.1% filtration rates for leukocyte, red cell, and hemoglobin, respectively. Contamination of concentrated genomic DNA was observed in the electrophoresis of RNA extracted from whole blood. Positive rates of hTERT, survivin and plakophilin were higher in the filtered MNCs. The filtration method detected 1 SNU484 cell/mL, and the blood samples of 4 (30.8%) lung cancer patients showed positive results for RT-PCR. CONCLUSIONS: For the detection of cancer cells in the blood, the filtration method was very efficient, and LunX and MAGE A1-6 genes would be useful for the detection of blood cancer cells in patients with lung cancer.