The results of pulmonary metastasectomy by a median sternotomy approach were retrospectively studied in 38 patients. This approach enabled the exploration of both thoracic cavities and, subsequently, the detection of other metastases. Such occult metastases were present in 11 of the 38 patients, four patients had unsuspected metastases at the contralateral side. There was no postoperative mortality, the morbidity associated with the median sternotomy and bilateral exploration was low. Pneumothorax occurred in three patients, pulmonary infection in four patients and a superficial wound infection in one patient. Hospital stay ranged from 4-25 days (mean 10 days). The five-year actuarial survival rate was 15 per cent. Resectability did not correlate significantly with long-term survival, neither did primary histology, disease-free interval or the presence of bilateral disease. The number of metastases was a highly significant predictor of survival (p = 0.0004).