Early clinical studies with gefitinib showed promising efficacy and mild toxicity in patients with advanced non-small-cell lung cancer (NSCLC). Thus, gefitinib was an ideal agent to evaluate in a maintenance setting in stage III disease. Untreated patients with stage III NSCLC, a performance score of 0 to 1, and adequate organ function were eligible. All patients received cisplatin 50 mg/m(2) on days 1 and 8 plus etoposide 50 mg/m(2) on days 1 to 5, every 28 days for two cycles with concurrent thoracic radiation (1.8- to 2-Gy fractions per day; total dose, 61 Gy) followed by three cycles of docetaxel 75 mg/m(2). Patients whose disease did not progress were randomly assigned to gefitinib 250 mg/d or placebo until disease progression, intolerable toxicity, or the end of 5 years. The planned sample size was 672 patients to confer power of 0.89 to detect a 33% increase over the expected median survival time of 21 months (one-sided P = .025, log-rank test). Random assignment was stratified by stage, histology, and measurable versus nonmeasurable disease. Enrollment began in July 2001. An unplanned interim analysis conducted in April 2005 rejected the alternative hypothesis of improved survival at the P = .0015 level for 243 randomly assigned patients. The study closed, and preliminary results were reported. Now, with a median follow-up time of 27 months, median survival time was 23 months for gefitinib (n = 118) and 35 months for placebo (n = 125; two-sided P = .013). The toxic death rate was 2% with gefitinib compared with 0% for placebo. In this unselected population, gefitinib did not improve survival. Decreased survival was a result of tumor progression and not gefitinib toxicity.