Accurate mediastinal lymph node dissection during thoracotomy is mandatory for staging and for adjuvant therapy in lung cancer. Pre-therapeutic staging for neoadjuvant therapy or for video assisted thoracoscopic resection of lung cancer is achieved usually by CT-scan and mediastinoscopy. However, these methods do not reach the accuracy of open nodal dissection. Therefore we developed a technique of radical video-assisted mediastinoscopic lymphadenectomy (VAMLA). This study was designed to show that VAMLA is feasible and that radicality of lymphadenectomy is comparable to the open procedure.In a prospective study all VAMLA procedures were registered and followed up in a database. Specimens of VAMLA were analysed by a single pathologist. Lymph nodes were counted and compared to open lymphadenectomy. The weight of the dissected tissue was documented. In patients receiving tumour resection subsequently to VAMLA, radicality of the previous mediastinoscopic dissection was controlled during thoracotomy.37 patients underwent video-assisted mediastinoscopy from June 1999 to April 2000. Mean duration of anaesthesia was 84.6 (SD 35.8) minutes.In 7 patients radical lymphadenectomy was not intended because of bulky nodal disease or benign disease. The remaining 30 patients underwent complete systematic nodal dissection as VAMLA.18 patients received tumour resection subsequently (12 right- and 6 left-sided thoracotomies). These thoracotomies allowed open re-dissection of 12 paratracheal regions, 10 of which were found free of lymphatic tissue. In two patients, 1 and 2 left over paratracheal nodes were counted respectively. 10/18 re-dissected subcarinal regions were found to be radically dissected by VAMLA. In 6 patients one single node and in the remaining 2 cases 5 and 8 nodes were found, respectively. However these counts also included nodes from the ipsilateral main bronchus. None of these nodes was positive for tumour.Average weight of the tissue that was harvested by VAMLA was 10.1 g (2.2-23.7, SD 6.3). An average number of 20.5 (6-60, SD 12.5) nodes per patient were counted in the specimens. This is comparable to our historical data from open lymphadenectomy.One palsy of the recurrent nerve in a patient with extensive preparation of the nerve and resection of 11 left-sided enlarged nodes was the only severe complication in this series.VAMLA seems to accomplish mediastinal nodal dissection comparable to open lymphadenectomy and supports video assisted surgery for lung cancer. In neoadjuvant setting a correct mediastinal N-staging is achieved.