In bronchial carcinoma the assessment of operability requires an accurate evaluation of the regional and mediastinal lymph nodes. For this, both mediastinoscopy and computed tomography are often used on a routine basis today. The present work considers the relative value of these two methods of investigation. Fifty-seven patients with bronchial carcinoma, in whom both investigations were carried out prior to surgery, were included in this prospective study. In the computed tomogram, lymph nodes with a diameter of more than 1 cm were defined as positive (i.e. suspected malignant infiltration). Sixteen of the 57 patients had histologically confirmed lymph node metastases; in 13 cases the metastases were detected by computed tomography, in 12 cases also by mediastinoscopy and in 3 cases only at thoracotomy. In 41 of the thoracotomized patients, no mediastinal metastases were found. As was to be expected, mediastinoscopy also proved negative in these cases. In 9 of these cases, however, the preoperative computed tomography findings were false-positive. For computed tomography the specificity was 78% and the sensitivity 81%; for mediastinoscopy, on the other hand, the specificity was 100% and the sensitivity 75%. The specificity of computed tomography is too low. Also, lymph nodes which are only inflamed may be considerably enlarged and cannot be differentiated in the computed tomogram from those with malignant infiltration. Mediastinal lymph nodes which appear enlarged in the computed tomogram therefore have to be further investigated by mediastinoscopy. However, in this prospective study, mediastinoscopy provided no additional information in those cases in whom the computed tomography findings were negative (lymph node diameter less than or equal to 1 cm); it did not detect the metastases in the 3 patients with false-negative computed tomography findings. Therefore, in the case of a negative computed tomogram, thoracotomy may be performed immediately, without previous mediastinoscopy.