STUDY OBJECTIVES: To review the etiology and pathophysiology of pneumomediastinum in severe blunt trauma, with a special interest in one of its possible origins, the Macklin effect. The Macklin effect relates to a three-step pathophysiologic process: blunt traumatic alveolar ruptures, air dissection along bronchovascular sheaths, and spreading of this blunt pulmonary interstitial emphysema into the mediastinum. The clinical relevance of the Macklin effect was also evaluated. SETTING: A university hospital serving as a reference trauma center. PATIENTS: A selection of 51 patients with severe blunt trauma between 1995 and 2000. Inclusion criteria: Severe trauma or high-speed deceleration justifying chest CT; if chest CT demonstrated a pneumomediastinum, bronchoscopy and esophagoscopy were performed to rule out tracheobronchial or esophageal injury. DESIGN: Retrospective analysis of patients' clinical files, chest CT, and bronchoscopy and esophagoscopy reports. The Macklin effect was diagnosed when an air collection adjacent to a bronchus and a pulmonary vessel could be clearly identified on the chest CT. Clinical relevance of the Macklin effect was statistically evaluated regarding its repercussions on the pulmonary gas exchange function, the respective durations of intensive care and total hospital stay, and the associated injuries. RESULTS: Twenty (39%) Macklin effects and 5 tracheobronchial injuries (10%) were identified. One tracheobronchial injury occurred simultaneously with the Macklin effect. The presence of the Macklin effect affected neither the clinical profile nor the result of pulmonary gas analysis on hospital admission, but was associated with a significant (p < 0.001) lengthening of the intensive care stay. CONCLUSIONS: The Macklin effect is present in 39% of severe blunt traumatic pneumomediastinum detected by CT. Its identification does not rule out a tracheobronchial injury. The Macklin effect reflects severe trauma, since it is associated with significantly prolonged intensive care stay.