In the scope of a prospective multi-centre study after neoadjuvant combined chemotherapy (carboplatin, ifosfamide, etoposide, vindesine) and radiotherapy (45 Gy) 40 resection specimens of locally advanced non-small-cell lung cancer were analysed in order to establish reproducible pathological/anatomical results of tumour regression. Resection specimens of 28 squamous cell carcinomas and 12 adenocarcinomas were investigated using serial sections of the primary lesion. The mean age of the patients was 57 years. The results were compared to spontaneous regressive changes in a control group of 50 untreated non-small-cell lung cancers. Marked scarry fibrosis in the region of the former primary tumour, concentric foci of fresh tumour necroses and surrounding foam cell clusters with transition into vascular granulation tissue could be established as characteristic features of therapy-induced tumour regression, whereas untreated carcinomas revealed necroses with adjoining vital tumour tissue. Using a three-step regression system, 3 tumours could be classified as grade I (no or only slight tumour regression), 10 tumours as grade IIA (marked but incomplete tumour regression, more than 10% vital tumour tissue), 20 tumours as grade IIB (less than 10% vital tumour tissue) and 7 tumours as grade III (complete tumour regression without vital tumour tissue). After a median follow-up period of 32.3 months in patients with grade IIB or III tumour regression ("responders") the median survival time of 27.9 months was found to be significantly longer than in patients with grade I or IIA tumour regression ("non-responders") with a median survival period of 13.7 months (log-rank test, P = 0.020). The resection specimens analysed, which were obtained 7 weeks (on average) after the end of radiochemotherapy, did not show specific changes due to preoperative therapy, but quite characteristic histological alterations in the former tumour area were registered, which had been induced by combined neoadjuvant radiation and chemotherapy. The grade of therapy-induced tumour regression could be shown to be a significant prognostic factor in non-small-cell lung cancer.