Previous investigations have suggested that elevated airway pressures increase the risk of ventilator-induced pneumothorax. However, risk factor analysis using multivariate techniques has not been done. We investigated the hypothesis that airway pressures would not independently correlate with pneumothorax when underlying disease was considered. All ventilated patients over a 1 yr period in the Hohenburg Critical Care Unit at the University of Alabama were followed until death or discharge from the ICU. Ventilator data were collected daily and the presence of pneumomediastinum and pneumothorax determined by review of chest radiographs. Maximal values of airway pressures, minute ventilation, tidal volume, and respiratory rate, as well as age, sex, and underlying disease, were entered into logistic regression analysis. A total of 168 patients was studied, and 20 experienced pneumothorax. Multivariate analysis of the entire ventilated population revealed that only the presence of ARDS independently correlated with pneumothorax. A similar analysis performed on the ARDS population revealed independent correlation only with male sex. Trends toward elevation in airway pressures were seen that did not reach statistical significance. We conclude that development of pneumothorax is most closely correlated with underlying disease, specifically ARDS, and that the associations previously noted between airway pressures and barotrauma largely relate to the occurrence of high airway pressures in ARDS.