Summary The role of post-operative radiotherapy for patients with non-small-cell lung cancer (NSCLC) is unclear despite five previous randomised trials. One deficiency with these trials was that they did not include adequate TNM staging, and so the present randomised trial was designed to compare surgery alone (S) with surgery plus post-operative radiotherapy (SR) in patients with pathologically staged TI-2, NI-2, MO NSCLC. Between July 1986 and October 1993, 308 patients (154 S, 154 SR) were entered from 16 centres in the UK. The median age of the patients was 62 years, 74% were male, >85% had normal or near normal levels of general condition, activity and breathlessness, 68% had squamous carcinoma, 52% had had a pneumonectomy, 63% had NI disease and 37% N2 disease. SR patients received 40 Gy in 15 fractions starting 4-6 weeks post-operatively. Overall there was no advantage to either group in terms of survival, although definite local recurrence and bony metastases appeared less frequently and later in the SR group. In a subgroup analysis, in the NI group no differences between the treatment groups were seen, but in the N2 group SR patients appeared to gain a one month survival advantage, delayed time to local recurrence and time to appearance of the bone metastases. There is, therefore, no clear indication for post-operative radiotherapy in Nl disease, but the question remains unresolved in N2 disease.Keywords: non-small-cell lung cancer; randomised trial; post-operative radiotherapy For patients with potentially resectable, non-small-cell lung cancer (NSCLC) without distant metastases, the standard treatment is an intended curative resection. In 1986, at the time this trial was planned and activated, a number of studies had already clarified some of the factors affecting resectability and subsequent prognosis.Firstly, it was already clear that the duration of subsequent survival was greatly reduced if the mediastinal nodes were found to be involved. For example, in a consecutive series of 245 patients undergoing curative resection, Wilkins et al. (1978) reported 5-year survival rates of 42% in patients without mediastinal node involvement compared with only 16% with such involvement. Greschuchna and Maassen (1980) reported corresponding rates of 37% and 11%, and broadly similar results were reported by others (Edwards, 1979;Mountain, 1986).Secondly, the importance of careful preoperative staging in the selection of patients suitable for intended resection was appreciated, particularly the need to avoid inappropriate thoracotomy in patients with T3 tumours, gross mediastinal node involvement or metastatic disease (Pearson, 1980;Spiro and Goldstraw, 1984). This need can only be met if, in patients otherwise suitable for thoracotomy, the mediastinum is staged, using such techniques as computerised tomography (CT), cervical mediastinoscopy and anterior mediastinotomy (if so indicated for left upper lobe tumours), and if clinical