This randomized study was designed to determine the response rates, survival and toxicities of single-agent gemcitabine (GEMZAR) and a combination of cisplatin/etoposide in chemonaive patients with non-resectable, locally advanced or metastatic non-small cell lung cancer (NSCLC). Gemcitabine 1000 mg/m2 was given as a 30-min intravenous infusion on days 1, 8, 15 of a 28-day cycle, cisplatin 100 mg/m2 on day 1, and etoposide 100 mg/m2 on days 1 (following cisplatin), 2 and 3. Major eligibility criteria included histologically confirmed non-small cell lung cancer, measurable disease, Zubrod performance status 0-2, no prior chemotherapy, no prior radiation of the measured lesion, and no CNS metastases. One hundred and forty-seven patients were enrolled, 72 in the gemcitabine and 75 in the cisplatin/etoposide arm. Patient characteristics were well-matched across both arms. Sixty-seven gemcitabine and 72 cisplatin/etoposide patients were qualified for efficacy analysis. There were no complete responses, but 12 partial responses in the gemcitabine arm and 11 in the cisplatin/etoposide arm, for protocol-qualified response but 12 partial responses in the gemcitabine arm and 11 in the cisplatin/etoposide arm, for protocol-qualified response rates of 17.9% (95%, CI: 9.6-29.2%,) and 15.3% (95% CI: 7.9-25.7%,), respectively. Median survival times were 6.6 months (95% CI: 4.9-7.3 months) for gemcitabine and 7.6 months (95% CI: 5.4-9.3 months) for cisplatin/etoposide. The 1-year survival probability estimate was 26% for gemcitabine and 24% for cisplatin/etoposide. There were no statistically significant between-group differences in time-to-event measures, but patients in the gemcitabine arm had a greater probability of achieving a tumour response after 2 months (probability estimate: 8 vs. 0%,) and of the response lasting at least 6 months (73 vs. 45%,). Clinical and haematologic toxicity was more pronounced in the cisplatin/etoposide arm. Quality-of-life measures indicated a significant worsening of symptomatology in the cisplatin/etoposide arm for hair loss, nausea and vomiting, and appetite loss. This randomized study provides further evidence that single-agent gemcitabine is an active and effective therapy for patients with non-resectable. locally advanced or metastatic NSCLC and good performance status, and that it is better tolerated than the combination cisplatin/ etoposide.