The Cancer and Leukemia Group B (CALGB) conducted a prospective randomized trial to evaluate the role of warfarin and alternating chemotherapy in extensive small-cell lung cancer (SCCL). After stratification for sex and performance status, patients were randomly assigned to receive chemotherapy with methotrexate, doxorubicin (Adriamycin; Adria Laboratories, Columbus, OH), cyclophosphamide, and lomustine (CCNU) (MACC), or MACC plus warfarin (MACC + W), or mitomycin, etoposide, cisplatin, and hexamethylmelamine alternating with MACC (MEPH/MACC). Warfarin was given continuously to maintain a prothrombin time of one and one half to twice the control values. A total of 328 patients were enrolled, and 294 were evaluable. There was a statistically significant advantage in objective response rates (complete [CR] and partial responses [PR], respectively) for MACC + W (17% and 50%) as compared with MACC alone (8% and 43%) or MEPH/MACC (10% and 38%) (P = .012). Both failure-free survival (P = .054 Wilcoxon test) and overall survival (P = .098 Wilcoxon test) were higher on MACC + W (median, 6.6 months and 9.3 months, respectively), as compared with MACC (5.0 months and 7.9 months) and MEPH/MACC (5.0 months and 7.9 months). Toxicity was comparable among the three arms, except for increased hemorrhagic events on MACC + W, which were life-threatening in four patients (4%), and lethal in two others (2%). These data support the role of warfarin in the treatment of SCCL, but do not establish its mechanism of action. Warfarin deserves further studies in SCCL, particularly in patients with limited disease.