We compared late thoracic radiotherapy (TRT) with early TRT in the treatment of limited-disease small-cell lung cancer (LD-SCLC). Patients with LD-SCLC received four cycles of etoposide plus cisplatin every 21 days. Patients were randomly assigned to receive either TRT administered concurrently with the first cycle (early TRT) or the third cycle (late TRT) of chemotherapy. The primary end point was complete response rate. Two hundred twenty-two patients were randomly assigned. Late TRT was not inferior to early TRT in terms of the complete response rate (early versus late; 36.0% versus 38.0%). Other efficacy measures including overall survival median, 24.1 versus 26.8 months; hazard ratio (HR) 0.90; 95% CI 0.18–1.62 and progression-free survival (median, 12.4 versus 11.2 months; HR 1.10; 95% CI 0.37–1.84) were not different between two arms. No statistical difference was noted in the pattern of treatment failures. However, neutropenic fever occurred more commonly in the early TRT arm than the late TRT arm (21.6% versus 10.2%; P = 0.02). In LD-SCLC treatment, TRT starting in the third cycle of chemotherapy seemed to be noninferior to early TRT, and had a more favorable profile with regard to neutropenic fever.