Endoscopic biopsy prior to chemotherapy provides an opportunity for studying biomarkers to predict the overall survival in gastric cancer patients. This prospective study was performed to identify prognostic biomarkers in patients with unresected gastric cancer. Fifty-nine cases of chemotherapy-naive metastatic gastric cancer were enrolled in this study. A microarray analysis was performed using 40 biopsy samples to identify candidate genes whose expressions might be correlated with the overall survival. After adjusting for clinical covariates based on a multivariate analysis, the identified genes were validated using real-time reverse transcription polymerase chain reaction (RT-PCR) analysis in 19 independent validation samples. Ninety-eight candidate genes whose expression levels were significantly correlated with the overall survival were identified using a microarray analysis based on a proportional hazards model (P < 0.005). Multivariate analysis was performed to assess 10 of these genes, and the results yielded a statistical significance level for DACH1 and PDCD6. We further evaluated these two genes in independent samples using real-time RT-PCR and found that lower mRNA expression levels of PDCD6 were correlated significantly with a poor overall survival. We identified PDCD6 as a prognostic biomarker in patients with unresected gastric cancer using endoscopic biopsy samples. Our PCR-based single gene prediction strategy successfully predicted the overall survival and may lead to a better understanding of this disease subgroup.