Two different simulation models were constructed to describe O2 diffusion into the bacteria-infected cells of legume nodules: one based on a central zone of uniform spherical cells and the other on a central zone of packed, uniform cubical cells with air spaces along the edges. The cubical model more closely approximated the geometry and gas diffusion characteristics of infected cells than did the spherical model. The models relied on set values for the innermost O2 concentration in the infected cell (1-20 nM) and predicted values for the free O2 and oxygenated leghemoglobin gradients toward the cell:space interface. The cubical model but not the spherical model predicted saturation of leghemoglobin (Lb) oxygenation at or within a few micrometers of the gas-filled intercellular space and predicted that the space concentration could be as high as 1.3% O2 when the fractional oxygenation of Lb and respiration rate within the infected cell were typical of that which has been measured in vivo. In the model, the higher the space O2 concentration, the greater the saturation of Lb by O2 and the greater the collapse of Lb-facilitated diffusion near the cell:space interface. This was predicted to result in a greater resistance to O2 diffusion from the space to the bacteroids, thereby providing an intrinsic, homeostatic mechanism for controlling the rate of O2 influx into infected cells. Changes in the physiological features of the simulated cubical infected cell, such as the proportion of the cell as cytosol, the surface area of the cell exposed to a space, the maximum rate of cellular respiration, or the concentration of Lb in the cytoplasm, significantly altered the extent to which the infected cell would be able to regulate its diffusive resistance. These results demonstrate the possibility of a Lb-based mechanism for controlling the O2 concentration within the infected cells. If such a mechanism exists in legume nodules, it would give the infected cell an ability to exercise fine control over its internal environment, a process that could complement a physical diffusion barrier that may exist in the inner cortex or elsewhere in the nodule and provide coarse control over O2 diffusion.