Surely no other surgical procedure has attracted the passion and sustained debate which has surrounded mediastinoscopy. Thoracic surgeons even now, 33 years after its introduction, are divided on the role of this investigation. Those who argue in favour of mediastinoscopy, and here I must declare my interest, believe it to be an essential staging investigation, to be considered in all cases of lung cancer prior to thoracotomy, providing valuable information as to whether pulmonary resection is (a) feasible and (b) desirable. Those who have not accepted mediastinoscopy look upon it as an invasive investigation, with its own morbidity and occasional mortality, which is difficult to interpre.t, will delay and complicate thoracotomy and which does not provide useful prognostic information. These surgeons may on occasions undertake mediastinoscopy to confirm mediastinal involvement and establish histological diagnosis where clinical or radiological features strongly suggest this possibility, but this is not the debate. The real issue is whether routine preoperative assessment of the mediastinum is necessary prior to thoracotomy for lung cancer, and whether mediastinal exploration by mediastinoscopy and mediastinotomy helps in this respect. I can only comment on this debate from the biased perspective of a protagonist of mediastinoscopy. I cannot hope to influence those who have established their own routine for the preoperative assessment of lung cancer, but will try to present to others the state of the art of mediastinoscopy in 1992. Cervical mediastinoscopy was first reported by Carlens in 1959 [l]. It was described as an endoscopic examination of the superior mediastinum through a short transverse cervical incision. The instrument is inserted beneath the pretracheal fascia, and dissection proceeds to the carina, allowing one to visualize and biopsy the lymph nodes in the right and left paratracheal regions, those in the pretracheal area and at the carina [2]. One can reach beneath the azygos vein to the nodes in the superior pole of the right hilum, and assess invasion into the right paratracheal area, and the left paratracheal area above the aortic arch. It will therefore be at its most sensitive in assessing mediastinal invasion and lymphatic spread of right upper lobe cancers [3]. When assessing tumours arising in the left upper lobe, or tumours which have reached the left main bronchus, cervical mediastinoscopy should be supplemented by left anterior mediastinotomy to assess invasion or node involvement outside the aortic arch and into the sub-aortic fossa and anterior mediastinum [4,5].