PURPOSE To test the concept that cisplatin dose-intensity is important in the treatment of non-small-cell lung cancer (NSCLC), the Southwest Oncology Group (SWOG) performed a randomized trial comparing standard-dose cisplatin (SDCP) 50 mg/m2 days 1 and 8 on a 28-day cycle for eight cycles, high-dose cisplatin (HDCP) 100 mg/m2 days 1 and 8 for four cycles, and high-dose cisplatin plus mitomycin (HDCP-M) 8 mg/m2 day 1. To isolate the effects of dose-intensity versus total dose, the planned cumulative cisplatin dose was 800 mg/m2 in each arm. PATIENTS AND METHODS Between July 1988 and April 1990, 356 patients were enrolled and 323 were eligible and assessable. All patients had metastatic, measurable disease, were chemotherapy-naive, and had a performance status (PS) of 0 to 2. RESULTS Confirmed complete plus partial response rates were SDCP, 12%; HDCP, 14%; and HDCP-M, 27% (P < .05). Complete responses were uncommon (HDCP, 3%; HDCP-M, 4%) and were observed only in the high-dose arms. Progressive disease occurred more frequently in the SDCP arm (57%) compared with HDCP (38%) or HDCP-M (34%) (P < .05). However, there were no significant differences in median survival times (SDCP, 6.9 months; HDCP, 5.3 months; HDCP-M, 7.2 months; P = .53). The mean delivered dose-intensity for cisplatin was significantly greater in the high-dose arms: HDCP 41 mg/m2/wk and HDCP-M 39 mg/m2/wk, versus SDCP 23 mg/m2/wk (P = .05). The high-dose arms resulted in an increased incidence of ototoxicity, emesis, and myelosuppression, but similar degrees of renal toxicity and neuropathy compared with SDCP. CONCLUSION This study does not confirm evidence of a steep clinical dose-response curve for cisplatin in NSCLC at the cisplatin dose-intensities achieved. The addition of mitomycin increases the response rate, but does not improve survival. Continued evaluation of new agents in this disease is warranted.