OBJECTIVE To describe our experience with median sternotomy for resection of lung metastases and to assess whether computer tomography (CT) accurately predicts the number and extent of lung metastases. METHODS Retrospective case record study. METHODS University hospital/Cancer Centre, The Netherlands. METHODS 78 patients with pulmonary metastases from various histological types of tumours who were operated on through a median sternotomy during the 10-year period January 1985-January 1995. METHODS Median sternotomy for resection of lung metastases with the intention to cure. Extension of the incision in case of extended disease. METHODS Presence of unilateral or bilateral metastases in relation to preoperative CT. RESULTS 78 patients underwent a total of 82 sternotomies. CT did not accurately diagnose the extent of disease in 38 patients (49%). In 72 cases metastases were excised. In 58 patients (81%) histological examination showed tumour-free margins microscopically. 36 patients had bilateral metastases. CT showed unilateral disease in 49 patients. 14 (29%) had bilateral involvement. 4 patients required lobectomy and in two patients anterolateral extension of the sternotomy was necessary. Eleven patients (15%) developed minor complications. There was no operative mortality. CONCLUSIONS Bilateral staging and finding of occult metastases, complete surgical clearance in a one stage procedure, and lower morbidity are the reasons that we suggest that median sternotomy is the procedure of choice of resection of pulmonary metastases. For eligible patients the choice of surgical approach should not be made conditional on the results of CT alone.