Combined modality therapy is of great importance in the management of small cell lung cancer. Randomized studies of the design chemotherapy with or without thoracic irradiation are required to demonstrate the impact of radiotherapy on rates of survival, local control and adverse effects. The method of meta-analysis allows one to analyse in a single study a set of different clinical trials of the same design. The first purpose of this paper is to briefly present the method, and the results of a meta-analysis on the role of thoracic irradiation combined with chemotherapy in the treatment of this disease. Thoracic irradiation added to chemotherapy results in a major decrease in local relapse (from 65% to 40% at 2 years), a modest increase in overall survival (from 16 to 22% at 2 years), and a small increase in lethal toxicity (from 1 to 2%). The second purpose is to discuss timing of thoracic irradiation with respect to the administration of chemotherapy using the results of a recently published trial. This Canadian study is based on the hypothesis that chemo resistant cells may develop as a result of spontaneous mutations during therapy. Limited disease patients all received the same chemotherapy (alternating 3-week cycles of CAV [cyclophosphamide, doxorubicin, vincristine] over EP [etoposide, cisplatinum] for a total of six cycles), after having been randomized to receive thoracic irradiation given either early (at 3 weeks following the beginning of treatment) or late (at 15 weeks). Of the 308 patients for analysis 155 were randomized to the "early radiotherapy" arm, and 157 to the "late radiotherapy". Prognostic factors were equally distributed between the two arms. The radiotherapy regiment consisted of a dose of 40 Gy in 15 fractions to a target volume including the primary tumor and mediastinum, and prophylactic brain irradiation (25 Gy in ten fractions in 2 weeks) to patients without progression of disease. The overall survival rate was better in the "early radiotherapy" arm, with a median survival of 21 months and on overall survival rate of 20% at 5 years, compared to a median of 16 months and a 5 years survival of 11% in the "late radiotherapy" arm. The survival curves are significantly different by the log rank (P = 0.008) and Wilcoxon (P = 0.005) tests, in favour of "early radiotherapy". After allowing for prognostic factors (sex, ECOG performance status) by the Cox model, the "early" arm retains a statistically significant advantage (P = 0.006).(ABSTRACT TRUNCATED AT 400 WORDS)