Non-small cell lung cancer (NSCLC) is a leading cause of death and a substantial fraction of patients with surgically resected disease ultimately dies due to distant metastasis. To identify gene expression differences in early stage adenocarcinoma that either did or did not metastasize within a 5-year period, we employed a subtractive hybridization strategy of pooled RNA from primary adenocarcinomas (stage I) of the lung. Individual clones (n=225) of the subtracted cDNA library were sequenced. Further analyses of mRNA expression levels in a cohort of 70 NSCLC patients (stage I to IIIA) showed that the metastasis association of the identified genes was stage and histology specific. Cox regression analyses identified two genes (EIF4A1, MALA1) to be independent prognostic parameters for patients' survival in stage I and II disease. These findings could help to identify early-stage NSCLC patients at high risk for the development of distant metastasis.