A study of 10 anaesthetised patients placed in the lateral position for thoracoscopic surgery assessed whether transient increases in pleural pressure on the side of the non-ventilated lung might increase the speed at which gas vents from that lung. The transient increases in pleural pressure were generated by the mediastinal displacement that occurs with each inspiratory phase of positive pressure ventilation of the dependent lung. When combined with a unidirectional valve allowing gas to flow out of the non-ventilated lung, and a second valve allowing ambient airflow into, but not out of, the thoracic cavity via an initial thoracoscopy access site, this mediastinal displacement could conceivably serve to 'pump' gas out of the non-ventilated lung. Using the four different combinations of valve inclusion or omission, the volume of gas that vented from the non-ventilated lung into a measuring spirometer was recorded during a 120-s measurement sequence. It was found that the speed of venting was not increased by the transient increases in pleural pressure, and that in all but one of a total of 34 measurement sequences, venting had ceased by the end of the sequence. Gas venting was a mean (SD) of 85.5 (11.9)% complete in 25 s (five breaths), and 96.6 (6.1)% complete in 60 s. This prompt partial lung collapse very likely reflected the passive elastic recoil of the lung, while the failure of transient increases in pleural pressure to result in ongoing venting of gas was probably a consequence of airways closure as the lung collapsed. It is concluded that techniques that aim to speed lung collapse by increasing pleural pressure are unlikely to be effective.