To evaluate the efficacy of a three-drug regimen vs. a two-drug CDDP based combination in the treatment of NSCLC, we conducted a three-arm randomized parallel trial comparing (a) CDDP (120 mg/m2 day 1) + etoposide (100 mg/m2 days 1-3) every 3 weeks (PE--arm A); (b) CDDP (120 mg/m2 every 4 weeks) + mitomycin (8 mg/m2 days 1, 29, 71) + vindesine (3 mg/m2 days 1, 8, 15, 22 every 2 weeks) (MVP--arm B); and (c) CDDP (120 mg/m2 day 1) + mitomycin (6 mg/m2 day 1) + ifosfamide (3 g/m2 day 2) every 3 weeks (MIC--arm C). From May 1989 to April 1992, 393 consecutive previously untreated patients with NSCLC Stage IIIB and IV entered the trial; 373 were evaluable for survival and 360 for response. The response rate was significantly better for both the three-drug regimens compared with PE (Table 3). Logistic regression model showed a significantly better response in patients with a good P.S. and in Stage IIIB. Main toxicity consisted of myelosuppression: neutropenia Grade III-IV was recorded in 14% (arm A), 15% (arm B) and 21% (arm C). Thrombocytopenia Grade III-IV was worst in arm C: 10% vs. 5% (arm A) and 3% (arm B). Nephrotoxicity Grade III-IV was more common in arm C: 3.5%. Toxic deaths were 11 (3%: three in arm A, five in arm B, three in arm C). From our data, the three-drug containing regimens, MVP and MIC, appear more active than the two-drug combination PE in treatment of advanced NSCLC.