Purpose Increased expression of eicosanoids in cancer has been associated with adverse prognosis. We performed a randomized phase II trial to test the hypothesis that inhibitors of two eicosanoid pathways (cyclooxygenase-2 [COX-2], celecoxib and 5-lipoxygenase [5-LOX], zileuton) added to chemotherapy would improve outcome in advanced non–small-cell lung cancer (NSCLC). Patients and Methods Patients with advanced NSCLC, a performance status of 0 to 2, and no prior therapy were eligible. All patients received carboplatin area under the curve (AUC) 5.5 mg/mL · min day 1 + gemcitabine (1,000 mg/m²) days 1 and 8. Patients were randomly assigned to: (a) zileuton 600 mg PO qid, (b) celecoxib 400 mg PO bid, or (c) celecoxib and zileuton at the same doses. Immunohistochemical staining for COX-2 and 5-LOX was performed without knowledge of outcomes. Results One hundred forty patients were entered and 134 were eligible and treated. There was no survival difference between the arms. COX-2 expression was a negative prognostic marker for overall survival (OS; hazard ratio [HR] = 2.51, P = .019 for index ≥ 4; HR = 4.16, P = .005 for index = 9) for patients not receiving celecoxib. Patients with increased COX-2 expression (index ≥ 4), receiving celecoxib had better survival than did COX-2–expressing patients not receiving drug (HR = .342, P = .005 for OS; HR = .294, P = .002 for failure-free survival). Multivariate analysis confirmed the interaction of COX-2 and celecoxib on survival. 5-LOX expression was neither prognostic nor predictive. Conclusion This study failed to demonstrate the value of dual eicosanoid inhibition or benefit from either agent alone in addition to chemotherapy. However, a prospectively defined subset analysis suggests an advantage for celecoxib and chemotherapy for patients with moderate to high COX-2 expression.