Fault zones are complex, and show considerable variability in both structure and the distribution of associated fault rocks within the fault core: the zone that localizes most strain and displacement. It is the fault-core gouge zone and associated slip surfaces which provide the cross-fault seal when permeable layers are juxtaposed. Predicting the sealing properties of fault gouge zones is difficult but often required when evaluating faults in exploration prospects. A stochastic modelling approach is described to help better understand the compositional controls on fault gouge seal potential. The model is populated with a random assemblage of four fault rock components: shale smears, shaly gouge, cataclastic gouge and low-strain host-rock lenses. Harmonic averaging of permeability and arithmetic averaging of V_shale are then used to upscale the properties, and to propose a simple permeability–V_shale model for fault rocks. Practical application of the model is discussed by developing an empirical link between standard well-log data and associated fault rock effective permeability. This new approach has the potential to offer a simple well-log-based fault seal model. The utility of the model is demonstrated with a case study, comparing the results to those generated using other published techniques.