Patients with advanced nonsmall-cell lung cancer (NSCLC), good performance status, and no prior chemotherapy were randomized to receive one of three regimens: intravenous vindesine (V) 3 mg/m2 every 2 weeks; V 3 mg/m2 weekly for 5 weeks, followed by a dose every 2 weeks plus mitomycin (VM) 20 mg/m2 day 1 and then 15 mg/m2 every 6 weeks; or V at the more intensive dose rate plus cisplatin (VC) 120 mg/m2 with forced diuresis on days 1 and 29 and then every 6 weeks. A total of 435 patients were enrolled in the trial, with 410 (94%) assessable for prognostic characteristics and survival. Among the 375 patients assessable for response, only 58 (15%) achieved objective response. Single-agent V every 2 weeks was inactive (response rate less than 1%), effectively acting as a no-treatment arm. Among assessable patients receiving VM, 33 (27%) responded; among patients receiving VC, 24 (19%) responded. There was no statistically significant survival difference among the treatment arms, with median survival among those treated with V 14.8 weeks, VM 20.4 weeks, and VC 24.7 weeks; VC achieved borderline significance (P = .06) compared with V. In a prognostic factor analysis, treatment was not a significant factor (P = .447) for survival. Thus, in this large multicenter trial, neither a high-dose cisplatin combination nor a noncisplatin regimen (VM) with a comparable response rate had a significant survival advantage over minimal chemotherapy. New approaches are needed in advanced NSCLC.