BACKGROUND--Postoperative lung injury is a recognised complication of thoracotomy for which there are few data regarding incidence and outcome. METHODS--In a case controlled study the notes of all adult patients who developed acute lung injury (ALI) or acute respiratory distress syndrome (ARDS) following thoracotomy between 1991 and 1994 were examined and classified according to the guidelines of the American Thoracic Society/European Respiratory Society for ALI/ARDS. The predictive value of a routine preoperative assessment and duration of anaesthesia in determining those patients most likely to develop ALI/ARDS was assessed. RESULTS--Between 1991 and 1994 231 lobectomies, 103 pneumonectomies, and 135 wedge resections and segmentectomies were performed. The overall incidence of lung injury was 5.1%; 17 patients developed ARDS (two survived) and seven developed ALI (five survived). There was no significant difference compared with case matched controls in preoperative spirometric values, arterial oxygen tension (PaO2), or duration of anaesthesia. None of these parameters was useful in predicting those patients most likely to develop lung injury. CONCLUSION--Lung injury after thoracotomy is associated with a high mortality. Conventional parameters for preoperative assessment do not predict those patients most likely to develop ALI/ARDS in these circumstances.