Cancers of the lung and bronchus claim over 160,000 lives per year in the United States alone, which is greater than the combined deaths from colon, breast, and prostate cancer. The major histological subtypes of lung cancer identified by light microscopy are small cell lung carcinoma (SCLC) and non-small cell lung carcinoma (NSCLC), with the latter including adenocarcinoma, (AD), squamous cell carcinoma (SCC), and large cell lung carcinoma (LCLC).  Despite our best attempts to morphologically classify lung cancer, we remain unable to predict the clinical behavior of this diverse disease, particularly in NSCLC. The ineffectiveness of current classification schemes to predict clinical outcome has drastic implications for the management of patients.  For example, while chemotherapy is effective in some patients with lung cancer, many receive no benefit while unfortunately incurring sigificant side effects due entirely to their treatment.  The primary hypothesis of my research is that through the use of new molecular technologies, more accurate information will be available to clinicians for managing patients with lung cancer.  To accomplish this goal I will address two specific tasks in the work described in the current proposal.  First, using a large amount of preliminary data generated using research-based genomic tests, I will develop a diagnostic test which can be administered in the routine practice of clinical medicine.  Second, using a large cohort of clinical samples, I will validate the diagnostic test’s usefulness in the clinical setting.  Specifically, I expect to reliably identify groups of tumors with different a risk of cancer-specific death, patterns of tumor spread, and response to therapy.