Air permeability in unstructured porous media can often be predicted directly from air-filled porosity based on one or two measurements of air permeability at different air-filled porosities using existing predictive models. In structured media and media that change their physical characteristics (bulk density, water content, organic matter content, etc.) with time, the relationship between air permeability and air-filled porosity is often very complex and consequently models for predicting air permeability in such media are not yet well established. We developed a simple model concept for predicting air permeability in porous media with variable structure. Predictions are based on water content, air-filled porosity, and dry bulk density and the model uses a very limited set of input data. The model is able to predict air permeability throughout the entire water content-bulk density domain. A set of four measurements of air permeability at different values of water content and dry bulk density is required as input. The model was tested against a set of air permeability data from six porous materials (a total of 236 measurements) representing a wide range of material types, organic matter contents, water contents, and dry bulk densities. Excellent agreement between measured and predicted air permeability values was found except at very low values of air permeability for some of the materials.