The European Lung Cancer Working Party (ELCWP) designed a 3-arm phase III randomised trial to determine the role of accelerated chemotherapy in extensive-disease (ED) small-cell lung cancer (SCLC). Eligible patients were randomised between the 3 following arms: (A) Standard chemotherapy with 6 courses of EVI (epirubicin 60 mg m(-2), vindesine 3 mg m(-2), ifosfamide 5 g m(-2); all drugs given on day 1 repeated every three weeks. (B) Accelerated chemotherapy with EVI administered every 2 weeks and GM-CSF support. (C) Accelerated chemotherapy with EVI and oral antibiotics (cotrimoxazole). Primary endpoint was survival. 233 eligible patients were randomised. Chemotherapy could be significantly accelerated in arm B with increased absolute dose-intensity. Best response rates, in the population of evaluable patients, were, respectively for arm A, B and C, 59%, 76% and 70%. The response rate was significantly higher in arm B in comparison to arm A (P = 0.04). There was, however, no survival difference with respective median duration and 2-year rate of 286 days and 5% for arm A, 264 days and 6% for arm B and 264 days and 6% for arm C. Severe thrombopenia occurred more frequently in arm B but without an increased rate of bleeding. Non-severe infections were more frequent in arm B and severe infections were less frequent in arm C. Our trial failed to demonstrate, in ED-SCLC, a survival benefit of chemotherapy acceleration by using GM-CSF support.