BACKGROUND Cisplatin-based chemotherapy is generally considered the most active treatment for advanced non-small-cell lung cancer. The combination of cisplatin and etoposide had for some time been the standard treatment at our center. Of the other active regimens, cisplatin in combination with mitomycin-C, vindesine or ifosfamide (MVP or MIC) showed the highest response rates. We decided to perform a comparative trial of the three 'best' regimens in order to define a possible standard regimen in advanced NSCLC. METHODS From May 1989 to April 1992, 393 consecutive, previously untreated NSCLC patients, stages IIIB and IV, were randomized to receive either cisplatin (120 mg/sqm day 1) + etoposide (100 mg/sqm days 1-3) every 3 weeks (PE) or cisplatin (120 mg/sqm every 4 weeks) + mitomycin-C (8 mg/sqm days 1-29-71) + vindesine (3 mg/sqm days 1-8-15-22) (MVP) or cisplatin (120 mg/sqm day 1) + mitomycin-C (6 mg/sqm day 1) + ifosfamide (3 mg/sqm day 2) every 3 weeks (MIC). Of these, 382 were evaluable for survival and 360 for response. RESULTS Response rates were statistically higher for both MIC (40%) and MVP (36%) than for the PE arm (23%). Survival estimates analyzed by the log-rank test showed a significant benefit (p < 0.04) for patients treated with three-drug regimens (MVP; MIC) as compared to those in the PE arm. The main toxicity was myelosuppression; thrombocytopenia WHO grade 3-4 was worse in the MIC arm; nephrotoxicity grade 3-4 was also more frequent in the MIC arm. CONCLUSIONS A three-drug cisplatin-based regimen (MVP; MIC) should be considered as reference treatment in NSCLC.