7181 Background: Gefitinib, an inhibitor of epidermal growth factor receptor-tyrosine kinase, has shown favorable anti-tumor activity to a subset of patients with advanced non-small cell lung cancer (NSCLC). However, gefitinib failed to significantly prolong survival in comparison to placebo in unselected population. Consequently, selection of patients who will benefit from this treatment is clinically important. In the previous study, we identified gefitinib-resistance related genes (GRRGs) by cDNA microarray analyses and reported that the responsiveness of individual patients with NSCLC to gefitinib could be predicted by expression of GRRGs. The goal of this study is to establish clinically adoptive prediction system of clinical outcome of NSCLC patients treated by gefitinib. Methods: Tumors from 30 NSCLC patients treated daily with 250mg of gefitinib were evaluated for the expression of GRRGs by quantitative RT-PCR. Gefitinib response score (GRS) reflecting gene expression was calculated based on weighted vote method and this numerical scoring system was validated by leave-one-out approach and independent test cases. Results: The sensitivity of 15 learning cases (five partial response (PR) cases and 10 progressive disease (PD) cases) were predicted by leave-one-out cross validation with 93.3% (14/15) accuracy. This numerical scoring system correctly predicted responses to the drug in all of nine additional test cases (three PR cases and six PD cases). Notably, this system also could predict two stable disease (SD) cases who successfully controlled by gefitinib more than four months as responder. On the other hand, four patients who had been judged as SD but showed progression of the disease within three months were judged as non-responders. Moreover, gefitinib-treated patients with good signature had a statistically significant improvement in time to progression, and survival compared with patients with bad signature (13.0 vs 1.7 months, P < .0001; 27.7 vs 6.6 months, P = .0007, respectively). Conclusions: Our results suggest that clinical outcome of NSCLC patients treated with gefitinib could be predicted by expression levels of GRRGs determined by quantitative RT-PCR analyses. No significant financial relationships to disclose.