The current study was designed to test the POSSUM (Physiological and Operative Severity Score for the enUmeration of Mortality and Morbidity) in a population of lung resection candidates, and to propose its use as an instrument of evaluation of surgical outcome and quality in thoracic surgery. 250 consecutive patients submitted to lung resection from 1993 through 1996 at our institution were prospectively evaluated. Two significant predictive models were than yielded by logistic regression analysis (model I: POSSUM alone; model II: Combining POSSUM Physiological Score with predicted postoperative FEV1 ) and compared with each other by means of ROC curves analysis. The study of the areas under the ROC curves showed that these models were equally predictive of postoperative complications (area of model I=0.66; area of model II=0.67). Both models showed no significant differences between predicted and observed morbidity (chi-square test p > 0.05). In particular, in model II there was perfect agreement between observed and predicted morbidity in the group of patients with a predicted morbidity above 60%. These results suggest that POSSUM may be appropriately used as a tool of surgical audit even in lung surgery.