SUMMARY This has been a study of 86 cases of non-tuberculous and non-traumatic spontaneous pneumothorax seen during the last five years. Such spontaneous pneumothoraces are a common clinical entity; recurrences are fairly frequent; non-expansion of the lung is more rare. When external violence and tuberculous disease, etc., are ruled out, the cause is considered to be the rupture of blebs or bullae. This was proved in 10 cases where surgical exploration was done for persistent collapse. A check-valve mechanism causing tension in such blebs is discussed. Illustrations demonstrating this mechanism are shown. These blebs and bullae, increasing in size under tension may cause respiratory embarrassment by encroaching on surrounding tissue, or may rupture due to interference with blood supply to the bleb wall, or by excessive intrableb pressures and produce pneumothorax. In most cases, where a bleb ruptures, the walls seal the fistula and the lung expands uneventfully with disappearance of the bleb. In a few cases, however, guy-wire intrapleural adhesions, fibrosis from pleural disease, a pleural membrane, or, in the majority of cases, a check-valve mechanism which maintains a continuous or intermittent fistula, keep the lung collapsed, producing persistent and even tension pneumothorax. Such pneumothoraces must be differentiated from tension cysts of the lung and giant tension bullae, and the dangers of needling these are emphasized. Treatment of spontaneous pneumothorax is also discussed emphasizing the limited indicators for poudrage, and the necessity for and the good results of exploratory thoracotomy in persistent pneumothoraces. It is the author's opinion that the cause of persistent and recurrent spontaneous pneumothorax lies in the bronchioles and alveoli in the great majority of cases. Attempts at pleural symphysis with blood, silver nitrate, etc., will usually not give good results. Surgery is the method of choice.