We studied the efficacy of cisplatin-based polychemotherapy for non-small-cell lung cancer. One hundred nineteen patients with adenocarcinoma or large cell carcinoma were randomized to receive cyclophosphamide, adriamycin, cisplatin and mitomycin C (CAPM) or mitomycin C, cytosine arabinoside and tegafur (MCT), and 48 patients with squamous cell carcinoma were randomized to receive cisplatin, adriamycin and peplomycin (PAP) or mitomycin C, cyclophosphamide, tespamine, toyomycin and tegafur (MCTTT). Radiation was given to the chest in patients with stage I-III disease. The response rates were CAPM, 34.5%; MCT, 13.1% (p less than 0.01) and PAP, 63.3%; MCTTT, 42.3%. A significant difference in response rate between the CAPM and MCT regimens was observed only in stage IV patients and not in stage I-III patients. The median survival was 9.5 months in the CAPM arm vs. 6.5 months in the MCT arm (p less than 0.007), and 8.5 months in the PAP arm vs. 6.5 months in the MCTTT arm. Improved median survival for the CAPM regimen was noted only in stage IV patients and not in stage I-III patients when compared to patients given the MCT regimen, respectively. Nausea and vomiting were significantly increased in patients with cisplatin-based polychemotherapy. Myelosuppression was more severe with the CAPM regimen than with the other chemotherapy regimens. We concluded that cisplatin-based polychemotherapy, CAPM and PAP therapy were of more benefit to patients with disseminated non-small-cell lung cancer than MCT and MCTTT therapy.