Preoperative chemotherapy has become an accepted treatment for stage IIIA (N2) non small-cell lung cancer (NSCLC). The majority of induction regimens employ cisplatin, although the importance of cis-platin dose in combination is unclear. A randomized trial was conducted to address whether higher pre-operative cisplatin doses result in improved survival and increased pathologic complete response in NSCLC. Patients with stage IIIA clinically enlarged and biopsy-proven N2 lesions were randomly assigned to receive either high-dose cisplatin (HDCP) (100 mg/m2) or moderate-dose cisplatin (MDCP) (50 mg/ m2) in combination with ifosfamide (3 g/m2) and mitomycin (6 mg/m2). Disease was restaged after 3 cycles, and those patients with response or stable disease underwent thoracotomy. From March 1993 to February 1997, 83 patients were randomized: 46 received HDCP, and 37 received MDCP. Clinical characteristics were well matched. Radiographic response rate was 59% for HDCP patients and 30% for MDCP patients (P = 0.01). Thoracotomy was performed in 71 patients (86%), 58 of whom had resectable disease. Complete resection rate was 61% in the HDCP group, and 51% in the MDCP group (P = 0.5). Postoperative mortality was 11%. Pathologic complete response was observed in one patient who received MDCP. Median survival in the HDCP and MDCP groups was 13 and 11 months, respectively (P = 0.3). In conclusion, higher radiographic response rate is observed in patients who receive HDCP, but this study fails to show any significant improvement in either overall survival or pathologic complete response in this group of patients.