The 02 permeability of legume root nodules is under physiological control; decreases in permeability are triggered by various forms of stress. Two linked mathematical models were used to explore several hypotheses concerning the physical nature of the variable diffusion barrier in nodules. Respiration and diffusion of dissolved 02 and oxygenated leghemoglobin were simulated for the nodule cortex and the nodule interior. Measured nodule permeabilities were shown to be inconsistent with the hypothesis that large numbers of air-filled pores penetrate the diffusion barrier. Changes in the affinity of leghemoglobin for 02 or in the rate of cytoplasmic streaming in diffusion barrier cells did not result in the large changes in 02 permeability reported for real nodules. The presence or absence, but not the thickness, of aqueous plugs in radial pores through the cortex was found to have a large effect on permeability. Flooding of intercellular spaces, either between layers of cells in the cortex or in the nodule interior, also caused large changes in simulated permeability. The unsteady-state 02 method for determining nodule permeability was tested using data generated by the model. The accuracy of the method was confirmed, provided that certain assumptions (full oxygenation of leghemoglobin under pure 02 and uniform conditions in the nodule interior) are met.A diffusion barrier in legume root nodules restricts 02 flux into the nodule and thereby protects nitrogenase from inactivation by 02 (21,22). The responses of nodules to such seemingly unrelated factors as drought, nitrate, and defoliation all apparently involve changes in the gas permeability of this barrier (8,10,14,17,23,24 Variable gas permeability of nodules might result from changes in (a) the number or size of gas-filled radial pores through the barrier, (b) the thickness of water plugs in radial pores, (c) the extent of flooding between layers of cells, or (d) flooding of airspaces in the central zone.Under the "open pore" and "variable plug thickness" hypotheses, changes in nodule gas permeability result from changes in radial intercellular pores through a layer of closely packed cells in the nodule cortex. The open pore hypothesis postulates a low-resistance pathway through radial air-filled pores. Because the diffusion coefficient of 02 in air is 1O4 times that in water, diffusion through even a small number of open pores could dominate total diffusion (which also includes diffusion through cells) across the barrier. The low solubility of 02 in water or cytoplasm further decreases the diffusion rate in cells relative to air. It has been suggested that changes in either the radius (19,25) or the number (1 1) of open pores could control permeability. Under the variable plug thickness hypothesis (1 1), each pore contains a plug of water, and permeability depends on the thickness of the plug.The "flooded cortex" and "flooded interior" hypotheses have not previously been formulated explicitly. The