A multimodal therapy concept for small-cell lung cancer, which for patients with established pretherapeutic homolateral lymph-node metastases (N2) prescribes induction chemotherapy with subsequent resection as well as supplemental chemo- and radiotherapy, provided the opportunity to evaluate histologically the radiological diagnoses "complete remission" and "partial remission" using resection specimens. In 17 patients a 75% to 100% reduction in tumor size was achieved according to radiological diagnosis. Predictions of "no evidence of disease" or "evidence of disease" were only correct in ten cases. In the remaining seven cases, histology showed the radiological findings to be incorrect. This gives a 77% sensitivity for radiological diagnosis with no specificity. Moreover, differentiation between therapy effect on the primary tumor and on the N2 metastases gives similar results: sensitivity 64% and 67% respectively, specificity 33% and 25% respectively. It is concluded that, particularly after the tumor responds well to therapy, radiological techniques are unsuitable for establishing a diagnosis of "no evidence of disease" or "evidence of disease" in small-cell lung cancer. This is because on the one hand the radiological methods available do not permit clear differentiation between vital tumor tissue and necrosis or fibrosis, while on the other hand groups of vital tumor cells beyond the resolution power of X-ray technology will escape detection.