Carbonyl reductase (CBR) is a cytosolic NADPH-dependent oxidoreductase metabolizing prostaglandins, steroids, quinines, and anthracycline antibiotics. Many experimental studies have shown that CBR plays important roles in the regulation of tumor progression, but clinical significance of CBR status remains unclear. Thus, we conducted a retrospective study on CBR mRNA expression in lung cancer. Tumor tissues obtained from 59 non -small-cell lung cancer patients were analyzed by quantitative real-time reverse transcription-PCR assay to reveal clinical significance of CBR expression. Angiogenesis was measured immunohistochemically as intratumoral microvessel density (IMVD) using anti-CD34 monoclonal antibody CD34-IMVD) and anti-CD105 monoclonal antibody (CD105-IMVD). CBR mRNA expression was significantly reduced along with progression of primary tumors (the mean CBR mRNA/ GAPDH mRNA, 3.288 Â Â 10 À2 for pT 1 , 1.628 Â Â 10 À2 for pT 2 , and 1.175 Â Â 10 À2 for pT 3-4 disease, respectively; P = 0.02). Moreover, CBR mRNA expression in tumor with nodal involvement seemed to be reduced as compared with that in tumor without nodal involvement (the mean CBR mRNA/GAPDH mRNA, 1.446 Â Â 10 À2 and 2.531 Â Â 10 À2 , respectively), whereas the difference did not reach a statistical significance (P = 0.09). The mean CD105-IMVD for CBR-high tumor was 59.2, which was significantly lower than that for CBR-low tumor (130.6, P = 0.02), whereas no significant difference between the mean CD34-IMVDs for CBR-high tumor and CBR-low tumor was found. The 5-year survival rate of CBR-high patients was 68.3%, significantly higher than that of CBR-low patients (36.5%; P = 0.03). A multivariate analysis confirmed that CBR-high expression was a significant factor to predict a favorable prognosis (P = 0.04; relative risk, 0.39; 95% confidence interval, 0.16-0.98). Expression of CBR mRNA was a significant prognostic factor in non -small-cell lung cancer and was inversely associated with tumor progression and angiogenesis. (Cancer Epidemiol Biomarkers Prev 2005;14(8):1972 -5)