Isolated dog lungs and lobes, and whole monkey lungs, were studied while suspended in air, and while immersed in a low density fluidised bed to simulate a pleural pressure gradient. Gas distribution was investigated during two minute cycles between transpulmonary pressures of −5 and +30 cm water (minimum volume, MV and total lung capacity, TLC respectively), by (1) injecting boluses of xenon-133 during inflation and scanning the lung at total lung capacity, and (2) monitoring the xenon content of gas leaving the lung during the subsequent deflation. With a bolus given at MV there was preferential distribution to the upper lung and the washout typically showed a rising “plateau” with a steeper terminal portion (phase IV). These features were exaggerated in the presence of a gradient of transpulmonary pressure when the bolus still went preferentially to the superior lung, even when the lung was inverted. Individual lobes also distributed boluses unevenly and gave washouts with rising plateaux and phase IVs, implying inhomogeneity which must also contribute to whole lung behaviour. A two-compartment mathematical model of the lung was used to demonstrate that washout plateaux could be level, or rise or fall steadily, depending upon the relative compliance of the two compartments. However, the terminal phase IV must always rise if the initial concentration of tracer is greater in the upper compartment. Macaque lungs showed similar features, but, when suspended in air, boluses were uniformly distributed and washout plateaux were relatively flat.