In the treatment of small cell lung cancer, carboplatin/etoposide/vincristine (CEV) is one of the most active regimens. In contrast, the etoposide/vincristine (EV) combination also has produced acceptable results in patients with extensive disease. To evaluate the efficacy and survival of patients treated with EV in comparison to those treated with more intensive CEV chemotherapy, a prospective, randomized, phase III trial was performed. The protocol for the treatment groups was as follows: treatment A (156 patients): carboplatin 300 mg/m2 day I, etoposide 140 mg/m2 days 1 through 3, and vincristine 1.4 mg/m2 days 1, 8, and 15; and treatment B (161 patients): etoposide 200 mg/m2 days 1 through 3 and vincristine 1.4 mg/m2 days 1 and 8. Chemotherapy cycles in each treatment arm were repeated every 4 weeks. Doses were reduced by 20% when hematologic or nonhematologic toxicity (grade 4) occurred. In all, 317 evaluable patients were treated. The overall response rate for patients treated with CEV was 79.8% compared with 59.8% for those treated with EV (P < .001). The median length of survival was 10 months for CEV-treated patients compared with 9 months for EV-treated patients (P = .19). Based on long-term survival rates, there was an advantage for the CEV-treated patients if they had good performance status, were younger than 60 years, had no distant metastases, and achieved a complete response to first-line therapy. We conclude that patients with poor prognostic factors (ie, poor performance status, multiple distant metastases, and less than partial response to the first cycle of chemotherapy) should appropriately be treated with the less aggressive two-drug combination chemotherapy. On the other hand, patients with good prognostic factors should be treated as aggressively as possible, and they will benefit from the more aggressive induction chemotherapy.